May 1994 SCSB# 380

RESEARCH-BASED SOIL TESTING INFORMATION
AND FERTILIZER RECOMMENDATIONS
FOR PEANUTS ON COASTAL PLAIN SOILS


Chapter 4
Critical Level: Definition and Usage in Interpretation


F. R. Cox

In a soil testing program, the concentrations of nutrients removed with a given extractant are determined. For each nutrient, field experiments are conducted on specific crops to determine the extractable nutrient concentration below which there will be a response to application of that nutrient. The concentration that indicates the division between responsive and non-responsive conditions is termed the “critical level.”

After defining a critical level thus, it would seem a simple matter to interpret a soil test by recommending fertilizer below that concentration and not recommending fertilizer above it. Unfortunately, that is not the case. There are a number of factors to consider.

The first factor that should be realized is that we are dealing with a system that has to be evaluated statistically. Since we use replicated trials and multiple observations, the critical level is a mean value. Given repeated experiments under exactly the same conditions, 50% of the tests would result in critical levels above the original and 50% would be below the original. In other words, there is a typical range in critical values that should be described by a normal distribution, rather than a single point.

The range in critical values will be broadened as conditions vary. There are a host of variable conditions in the categories of soil, management, and climate. Examples of the effect of climate may be shown in recent work conducted in North Carolina. In one experiment, corn, soybeans, and wheat were grown during a nine-year period on a soil with a wide range in Mehlich-3 extractable P (M3P) (Cox 1992). With some double cropping, there were three to five observations for each crop. As expected, there were some differences in M3P critical level among crops, but also there were marked differences from year to year with the same crop. With the linear plateau method, mean    () M3P critical level across crops and years was 30 mg L-1 with a sample standard deviation (s) of 7 mg L-1. Thus, the sample standard deviation was about 1/4 of the mean. In a normal distribution, x s takes in 67% of expected observations and x 2s takes in 95%. With this information, a range in critical values could be given for this soil depending on the degree of inclusion desired.

A similar example may be given for peanuts grown by the author on a Goldsboro soil (fine-loamy, siliceous thermic Aquic Paleudult) at the Peanut Belt Research Station in North Carolina. Critical levels of P and K for the Mehlich-3 extractant were determined by means of the linear plateau method (Figures 1 and 2). Residual effects of prior fertilization were measured, and yield responses were noted to P after five years and to K after seven years. The M3P critical levels for three crops were 20, 25, and 17 mg L-1. This averages to 21 mg L-1. Mehlich-3 removes almost twice as much P as Mehlich-1, the extractant used in several other peanut-producing states, so with conversion to a weight basis the average M1P critical level would be about 8 mg kg-1. If the sample standard deviation is similar to that in the prior study, the MlP critical level range would be 6 to 10 or 4 to 12 mg kg-1depending on the degree of inclusion of expected observations desired ( ± s or 2s ).

Figure 1. Effect of Mehlich-3 P on the yield of three crops of peanuts grown on a Goldsboro soil in North Carolina. Critical levels are identified with a linear plateau.
 

The M3K critical levels were 0.12 and 0.09 cmol L-1 (Fig. 2), which average 0.105 cmol L-1. This value would be 32 mg K kg-1 with an assumed sample density of 1.3 g cm-3. As Mehlich-3 and Mehlich-1 remove similar amounts of K, the M1K critical level range could be 24 to 40 or 16 to 48 mg kg-1 depending on the degree of inclusion of expected observations desired.

Figure 2. Effect of Mehlich-3 K on the yield of two crops of peanuts grown on a Goldsboro soil in North Carolina. Critical levels are identified with a linear plateau (0.1 cmol L-1 = 39 mg L-1).
 


These examples suggest creation of a range in critical levels by differences in annual climatic conditions. There are also differences in soil and management factors that would decrease the precision of the critical level range if they are not taken into account. For instance, the range in P critical level decreases with increasing clay content (Cox and Lins 1984). This may not be an important factor when growing peanuts as the crop is ordinarily grown on sandy, low-clay sites.

The nutrient content of the subsoil also affects the amount of that nutrient required from the topsoil to meet plant needs. Woodruff and Parks (1980) found this especially true for K. If K fertilization is routinely greater than K removal, the subsoil would have a substantial reserve of available K and the critical level in the surface soil could still be rather low.

Disregarding soil and management factors should expand the critical level ranges in soil test interpretation. Similar results can be achieved by combining data from numerous sites, in which case the data are often transformed to “relative yield.” This approach is used frequently in soil test interpretation studies. The range would be the same, however, whether using actual or relative yields.

When a critical level range has been established for a crop, points within that range vary in probability of getting a response to fertilization. At the low end of the range, a yield response is highly probable and should occur almost 100% of the time. On the other hand, at the high end of the range, responses would seldom occur. This range is represented by the “medium” class in many soil test evaluation schemes. It covers the variable response range, whereas in the “low” class responses are always expected and in the “high” class responses are never expected.

The examples cited above are based upon interpreting the critical level range with the linear plateau method. When the quadratic plateau technique was applied to the peanut data, critical level ranges were 25 to 30% greater. When an exponential function at 95% of maximum yield was compared to the linear plateau with the three-crop data, the M3P critical level range was 67% greater with the former. The method of interpretation, therefore, may markedly affect the critical level range determined and should be made known.

References

COX, F. R. 1992. Range in soil phosphorus critical levels with time. Soil Sci. Soc. Am. J. 56:1504-1509.

COX, F. R., AND I. D. G. LINS. 1984. A phosphorus soil test interpretation for corn grown on acid soils varyng in crystalline clay content. Commun. Soil Sci. Plant Anal. 15:1481-1491.

WOODRUFF, J. R., AND C. L. PARKS. 1980. Topsoil and subsoil potassium calibration with leaf potassium for fertility rating. Agron. J. 72:392-396.


Document Prepared by:
Leigh H. Stribling, lstribli@acesag.auburn.edu
Alabama Agricultural Experiment Station
Auburn University

 return to top of page

  return to contents