Developing a soil test-crop response calibration
A soil test-crop response calibration for a specific crop type (e.g. wheat) is the relationship between the measured yield increase to an applied nutrient at a range of experimental sites and the soil test value for each trial site.
Researchers have commonly used the percent relative yield (% RY) (either as grain yield or crop biomass yield at flowering or crop maturity) as the estimate of yield responsiveness, largely because this estimate partially circumvents variations in yield responses to applied nutrient between sites and to variations in seasonal conditions that could affect grain yield and/or yield responsiveness to an applied nutrient.
% RY = (Yield of unfertilised treatment (Y0) / maximum yield obtained (Ymax)) x 100
Note: each trial confers a single point to the calibration relationship and non-responsive trials are as important as those that have responded to an applied nutrient.
The critical soil test level (and the variance, called the critical range) is defined at near maximum relative yield. The BFDC Interrogator determines the critical soil test criterion at 80, 90 and 95% RY.
These trial data can be partitioned to better define the critical soil test level according to:
- Soil types or soil texture classes.
- Crop descriptors (e.g. crop type).
- Environmental characteristics (e.g. growing season rainfall).
In some cases researchers have used yield increase to applied nutrient (Ymax - Y0) rather than % RY.
Estimating Y0 and Ymax
Several functions have been used to statistically fit relationships between crop yield (e.g. t/ha) and the level of applied nutrient (e.g. kg/ha). These functions can be used to estimate Y0 and Ymax using coefficients of the functions.
Note: all coefficients for these functions have positive values.
| Function | Equation | Fitted Y0 | Fitted Ymax |
|---|---|---|---|
| Mitscherlich | Y = A [1 - Be-CX] | A(1 - B) | A |
| Quadratic | Y = A + BX - CX2 | A | A + B2/4C |
| Square root | Y = A - BX + CX1/2 | A | A + C2/4B |
| Parabola | Y= A[1 - B[ X/D]2 - 1]2 | A(1 - B) | A |
Where Y = grain yield (t/ha); X = level of nutrient applied (kg/ha).
Deriving the soil test-crop response calibration
Each calibration curve was obtained for % RY by soil test value for a selection of trials. This was undertaken by fitting a straight line regression in the transformed domain of y = ln (soil test), x = arcsin (sqrt RY), computing the critical levels and ranges, then back transforming the regression line for presentation with the input data.
Given that the regression estimates the natural log (ln) of the soil test value that corresponds to a given critical level (e.g. RY = 90%), with appropriate confidence limits, that estimate has to be sensibly within range or involving little extrapolation.
top
