Soil test levels to guide nutrient stewardship: Phosphorus and potassium focus for the North-Central region

In what is the most comprehensive assessment of soil fertility status of North America, the 2020 North American Soil Test Summary is the fifth summary conducted since 2001. The effort was initiated by the International Plant Nutrition Institute (previously Potash & Phosphate Institute) with the 2020 summary presented by The Fertilizer Institute, Foundation for Agronomic Research, Plant Nutrition Canada, and PAQ Interactive. Interactive exploration of the 2020 and past summaries can be done at https://soiltest.tfi.org/. This article discusses trends and relationships of soil test levels for phosphorus and potassium for the conterminous 48 states (CONUS) and the North-Central (NC) region of the U.S.

Soil Testing—A Nutrient Management Tool

Soil testing provides a foundation for nutrient management decisions. Those decisions are generally directed towards achieving agronomic, environmental, and economic outcomes within the framework of agricultural production and land management. Conceptually, Figure 1 displays how system metrics such as crop yield, losses with surface runoff, and the probability of an economic response to fertilization all relate to a given soil test P (STP) level. From these relationships, one can see that assessing trends in soil test levels (across wide geographies) can inform general discussions around crop production, environmental concerns, and economic resiliency. Agronomic objectives, specifically those related to crop nutrients, can be achieved through a well-planned, executed, and interpreted soil-testing program.

Methods for soil sample collection vary and have been well documented (Peck & Soltanpour, 1990). Laboratory analyses used for soil samples destined to inform nutrient management also vary greatly by region and are commonly developed and adopted to fit specific soil properties and wet chemistry that provides consistent results and recommendations (Melsted, 1967). Interpreting soil test results bring all of the field and laboratory work together to inform agronomists and farmers about how the results relate to predictable crop responses to fertilization, profitable economic fertilization strategies, and identification of soils that require specific management practices. Soil test interpretation methods vary greatly by state and represent considerable underlying investigation and resources to connect data with recommended practices.

Soil testing for agronomic deliverables represents one of, if not the, largest opportunity to relate nutrient management decisions to environmental outcomes. For example, many states that have adopted a Phosphorus Index, designed to identify fields that would be susceptible to significant P losses, use agronomic-oriented sample depths and tests for use of the tools. It’s been well established that agronomic soil tests can be used for environmental quality metrics (Sims, 1998) and certainly have value in monitoring beyond their purely agronomic benefit. The economic implications of soil testing are many; however, they primarily reside with decisions to amend or fertilize (Dahnke & Olson, 1990). Beyond nutrient recommendations, soil testing can inform other components of cropping systems such as planting densities, irrigation scheduling, and others that certainly affect the bottom line.

Soil Test Summary

A similar protocol was used for the 2020 summary as was used for the previous efforts from 2001 to 2015 (TFI, 2021b). Forty-five private and public laboratories voluntarily submitted data for the 2020 summary. Soil test phosphorus (P), potassium (K), sulfur (S), magnesium (Mg), zinc (Zn), chloride (Cl^–), and soil acidity (pH) were included in the summary. Soil organic matter (SOM) was included for the first time in 2020. The data contributed to the summary were solely production farm soil samples, reported at the state level, with an identification for manured vs. non-manured and grid/zone vs. whole-field samples. For the purpose of this article, we will focus on P and K.

Soil Test Level Trends

Phosphorus and K are two macronutrients for which fertility recommendations largely rely on soil test levels. Not surprisingly, these nutrients have also been largely the focus of soil test correlation and calibration work to create specific recommendations for a given geography. The nature of both P and K in the bulk soil and rhizosphere and their diffusive nature in moving through different media drive the ability for the soil test to be useful for nutrient management. The 2020 summary included 7.3 million samples for P and 6.9 million samples for K representing the U.S. (Table 1.). While all P data are discussed as Bray-P1 equivalent values, it’s important to acknowledge that the use of specific soil P tests does vary by region. Regarding P, the Mehlich-3 test determined with inductively coupled plasma spectroscopy (ICP) was the most frequently reported test at 56% of the total P samples, followed by Bray-P1 (24%), Olsen (9%), Mehlich-1 (6%), and Mehlich-3 determined colorimetrically (4%). While a diversity of K tests certainly exist, the majority of samples reported (86%) were either Mehlich-3 or ammonium acetate.

Table 1 shows the number of soil samples represented by the 2001–2020 summaries for the CONUS, NC region, and its states included. Summed across the five summaries, about 23.7, 22.7, and 22.8 million samples represent P, K, and soil pH. Sample volume in the NC region mirrors CONUS trends. Sample volume between states in the NC region varies greatly and largely reflects row crop acres for each specific state. Significantly large increases in sample volume from each NC State occurred from 2001 to 2002 with the largest amount of samples for P, K, pH, and SOM submitted from Iowa, which showed more than 1 million samples for P, K, and pH. These data help to communicate just how large the farming and agronomy community investment has been in soil sampling as a tool used by farmers and agronomists.

The NC region of the U.S. as discussed here includes Iowa, Illinois, Indiana, Kansas, Michigan, Minnesota, Missouri, Nebraska, North Dakota, Ohio, South Dakota, and Wisconsin. This scaled-down, regional focus allowed for analysis of areas where similar dominant crops are grown, cropping systems share similar characteristics, and soil-test-based recommendation systems are widely used. The NC region also has made up about two-thirds of the U.S. soil samples submitted to the Soil Test Summary from 2001 to 2020 (Table 1.).

National data in Table 2 display median P, K, pH, and SOM for the CONUS, NC region, and its included states individually from 2001 to 2020. It is important to note the differences in sample volume representation between each summary year (previously discussed and shown in Table 1.) when interpreting changes in median values for a given region. Median STP for the CONUS has fluctuated from 2001 to 2020 with a general decreasing trend. From 2001 to 2020, the CONUS median STP dropped from 27 to 23 ppm P. Median soil test K (STK) changed from 154 in 2001 to 141 ppm K in 2020 and peaked in 2005 at 155 ppm K. Soil pH showed no change when median values across the entire CONUS are assessed (Table 2). The large diversity of regions across the CONUS makes changes across the broad geography difficult to capture, and realistically, local changes in acidity and soil pH should be addressed on a finer scale such as that done in Montana (Jones et al., 2019). Soil organic matter was first included in the 2020 summary, and so no trend assessment can be done yet, but as an important soil test measurement, it is worth noting the median SOM for the CONUS in 2020 was 2.9%.

State-specific soil test levels varied greatly and reflect the diversity of both management and natural soil test level effects over time. Individual states also varied in their trends from 2001–2020 (Table 2). For example, Illinois, Michigan, and Wisconsin all saw reductions in median STP over 10 ppm P, and states such as North Dakota and South Dakota increased 3 and 5 ppm P, respectively. Median STP stayed relatively steady in Iowa, Kansas, and Nebraska throughout all five summaries from 2001–2020 even with large increases in sample volumes. Median STK did not change over time in the same way STP did for many states and the NC region. States that showed the largest decreases were Kansas, Nebraska, and South Dakota with reduction of median STP of 120, 88, and 66 ppm K, respectively. Median STK increased in Iowa the most with a change of 27 ppm K from 2001 to 2020 (Table 2). Regionally, soil pH showed little change over time. Most states did not see large changes in soil pH as well, except for Kansas and South Dakota, which reported reductions of 0.5 and 0.3 pH units from 2001 to 2020. Median SOM was 3.0% for the NC region in 2020. State-specific median SOM values reflect the range of local parent materials, moisture regimes, and degrees of weathering and soil formation in the region.

To break down what is represented by a median soil test value change, Figure 2 shows the data for the CONUS and NC region segregated into soil test ranges. The y-axis in Figure 2 represents the change in the percentage of all samples existing in each range. For example, a greater percentage of samples for STP tested in the 0–5, 6–10, 11–15, and 16–20 ppm P ranges in 2020 compared with 2001 for both the CONUS and NC region. Values below 0, such as the STP range of > 50 ppm P, indicate a lower percentage of samples being in that category in 2020 than 2001. For STP, the trend of more samples testing below 20 ppm P and fewer samples testing above 20 ppm P matches well with the slight decreases in median STP shown in Table 2. Interestingly as well is that below 16–20 ppm P is where most soils would be considered “suboptimal” related to maintain STP for optimized agronomic function below which yield responses to fertilization would be expected. Distribution of STP for both the CONUS and NC region showed similar patterns; however, STP in the NC region tested more frequently in the 11–20 ppm P range and less frequently in the > 50 ppm P range when compared with the national data (Figure 2).

Soil test K distributions for the CONUS and NC region differed much more than for STP (Figure 2). While across the CONUS, STK increased significantly in the 0–120 ppm K ranges, NC data suggested very small changes from 2001 to 2020 below 120 ppm K. The NC region saw an increase in the percent of samples between 121–280 ppm K and no change from 281–300 ppm K, whereas CONUS data suggested there was either no change or reductions in the percent between 161–320 ppm K. Both nationally and regionally, STK was less frequently above 320 ppm K. As native STK and soil characteristics such as mineralogy change greatly between NC region states and the rest of the U.S., these differences in STK distribution are reasonable.

Soil Test Levels and Nutrient Use Data

The next step in this analysis was to relate soil test changes with nutrient use data from the NuGIS (Nutrient Use Geographic Information System) effort (TFI, 2021a). NuGIS aggregates fertilizer, recoverable manure nutrients, and crop removal with harvest at the county level and HUC8 watershed level for the CONUS (https://nugis.tfi.org/). Investigating these relationships allows for some reasoning behind changing soil test level and for identifying how trends in nutrient use dynamics affect the soil fertility status. Data for each state, region, and the CONUS for four years prior to each Soil Test Summary was compared with the soil test level of a given year. In other words, the average fertilizer, manure, and crop removal per acre per year for a given state from 1996–2000 was paired with the 2001 summary data, the 2001–2004 nutrient use data was paired with the 2005 summary data, and so on. Figure 3 shows fertilizer, manure, crop removal, and median soil test levels for P and K in the entire NC region. The stacked blue and red bars indicate P and K inputs, and the green bar represents crop removal with harvest. The difference between the stacked inputs and crop removal equates to the nutrient balance per acre per year (expressed as P₂O₅ and K₂O) (Figure 3).

Nutrient balances for both P and K were negative for each year in the NC region, meaning more P and K was removed with the crop than applied with fertilizer and/or manure (Figure 3). The relationship between balance and soil test level was very different for P and K. Median STP showed a decreased trend from 2001 to 2020 and fit a linear decrease well. Potassium did not show a strong relationship with K₂O balance per acre, as median STK showed a slight decreasing trend, even though K balance was between 12 and 16 lb K₂O ac^–1 yr^–1. Across such a large geography, relating the balance of nutrients to soil test level is difficult, and correlation should not be interpreted as causation. Median soil test levels for a single state at best can point to trends in soil fertility status over time if enough data is considered. What can be interpreted from Figure 3 is that at the same time negative P balances occurred, so too did the median STP decrease in the same region with paired data on the state level. Median STK did not correlate well with K balance, and as seen in Table 2, is much more difficult to assign a trend to than STP. The dynamic nature of K in soils and our cropping system could be the reasoning behind this.

Crop removal of nutrients with harvest informs many soil test recommendation systems for P and K across the county and specifically in the NC region. Increasing crop removal of P and K (Figure 4) from 1996 to 2016 is shown with trends in median STP and STK. Though across a broad area, these trends are important to consider when fertilization decisions are being made, particularly if recent soil test data are not available. Increases of about 10 lb P₂O₅ and K₂O ac^–1 yr^–1 were observed from within two decades, eluding to changes in removal rates possibly due to higher yield levels in the region or shifts towards crops that remove greater amounts of nutrients. Commonly linked in conversations for soil testing the fertilization decisions, P and K are not necessarily being removed at the same rate every year. Crop removal coefficients (pounds of nutrient removed per harvest unit of crop) vary greatly between crops, and as seen in Figure 5, NC region crops across the whole area are removing much more K than P. With dominate crops like corn and soybean, this should come as to no surprise if you’ve had to calculate removal rates in the past. However, when seeing the data, it’s a great reminder about the ratios of nutrients being removed from cropland acres on this large of a scale.

Implications for Nutrient Management