Is your precision ag strategy precise enough?

For decades, the primary concepts of modern soil fertility management have focused on utilizing semi-dense soil sampling (1 to 5 acres per sample) to obtain spatial soil test results, utilize some form of interpolation to estimate soil test levels between soil samples, and invoke any number of fertility recommendation calculations and algorithms to generate a fertilizer application map.  Much software has been built and potentially millions of acres of fertilizer have been applied in this manner with the goal of applying the correct fertilizer to the right parts of fields.  This has been the cornerstone of “Precision Agriculture.”

While this approach is relatively simple (determining sample placement on a grid using a defined extent), it doesn’t incorporate useful characteristics of the field, including soil water holding capacity, drainage and parent material.

SWAT MAPS provides the process to create SWAT zones, a much-improved way to represent the soil characteristics of a field.  SWAT stands for “Soil, Water and Topography”.  Zones created using the SWAT process incorporate the natural terrain of the field and help break areas down into common elements of similar soil characteristics that have similar topographical and water influence.  These zones can be used to place samples or sub-samples, allowing for better potential correlation between soil test values and the underlying soil properties.

In recently published research entitled, “Soil test phosphorus predicts field-level but not subfield-level corn yield response,” scientists set out to answer if, at a sub-field level, a recommended amount of phosphorus fertilizer showed a yield response in corn when the soil test phosphorus level was below the determined Critical Value.  For soil fertility recommendations, the Critical Value is a soil test level of a nutrient above which provides little to no economic benefit from additional fertilization, and below which has been shown to have economic response to fertilization.

More than 100 plots were placed across the landscape each at two field-sites in Kentucky, with each plot containing subplots with either a control (no phosphorus) or a treatment (with 29.5 kg/ha or 26.3 lbs/ac of P2O5 nutrient).  Each plot was either 9 x 9 m or 12.2 x 12.2 m (29.5 x 29.5 feet or 40 x 40 feet).  Applications of liquid ammonium polyphosphate with the planter in a subsurface band were applied in the no-till planting program.  This placement was utilized as it offered the highest probability of yield response.  These plots were harvested using a plot combine, harvesting the middle two of the four rows of each plot and yield captured by a yield monitor.  Soil samples were taken in each plot prior to each of the corn cropping seasons and analyzed using the Mehlich 3-ICP methods.  Seven site-years were captured using this process between 2016 and 2021.

Across all site-years, Soil Test P ranged from 1 to 63 ppm M3P and averaged 14 and 12 ppm for each site, respectively.  94 to 98% of the plots had M3P values below the 30 ppm critical value set by University of Kentucky guidance. 

Interestingly, soil pH was also low at both sites; one site averaged 4.6 pH and ranged from 4.0 to 7.3 in the first two cropping seasons before being limed and raised to 5.2 (range of 4.6 to 7.2).  The other site started with an average pH of 5.9 and ended at 5.3 at the conclusion of the study.

Comparing the yield between the untreated and treated subplots, it was found that 31 to 70% of the plots per site-year responded to phosphorus; on average, 51% of the plots responded across all site-years, and 96% of the plots had soil test P levels of less than 30 ppm, the critical value.

At the field scale, 5 of the 7 site-years did show that phosphorus application was effective enough to raise yield significantly.  However, at the plot level, this was not always observed.  In fact, some negative responses to phosphorus application were observed at the plot level.

These results indicate that while current soil test recommendations may be useful at the field level, they may not be so useful at the sub-field level.  Unfortunately, the sub-field level of management is what is considered as “Precision Agriculture,” the state of the art.
Does this suggest that soil sampling should be engaged at even higher resolutions to capture even greater sub-field variability?  Work by Lauzon et al., (2005) included the sampling of 23 fields in Ontario at 30 meter (96 feet or 0.21 acres/sample) spacing.  Using autocorrelation analysis, they found that 13 of the 23 fields would require sampling of 30-meter or less to adequately address their spatial variability.  At only one field was it found that the typical 100-meter grid (2.5 acres/sample) pattern was adequate for phosphorus and potassium assessment.  This is not a ringing endorsement of the current generally accepted approaches used in Precision Agriculture.

On the other hand, would field-level composite samples be the way to overcome the sub-field variability question?  This would inevitably mask portions of the field that may respond differently.  Ignorance is bliss.

Instead, the authors of the paper suggest future research should consider “incorporating mechanistic factors such as soil texture, climate zones, and crop production history;” in other words, incorporating sub-field “context” into corn yield response algorithms.  This would point towards utilizing tools that can parse out the variability of the field’s soil composition, water movement characteristics, and topography, like SWAT MAPS.

This research work was very comprehensive, but it would be worth considering at other locations with varying soil and weather conditions to see if these results hold true.  If so, this work may lead us towards a better “Precision Agriculture,” one that doesn’t hold a blind eye to the conditions of the soil but instead welcomes the natural (and human-influenced) variability to help explain yield response to nutrients.

Vaughn Reed, Jenni Fridgen, Bronc Finch, John Spargo, Josh McGrath, James M. Bowen, Gene Hahn, Douglas Smith, Edwin Ritchey, 2025.  Soil test phosphorus predicts field-level but not subfield-level corn yield response, Agronomy Journal, Volume 117, Issue 1

John D. Lauzon, Ivan P. O’Halloran, David J. Fallow, A. Peter von Bertoldi, Doug Aspinall. 2005. Spatial Variability of Soil Test Phosphorus, Potassium and pH on Ontario Soils.  Agronomy Journal, Volume 97, Issue 2.