FIELD-SCALE FERTILIZER RECOMMENDATIONS - THE SPATIAL SCALING PROBLEM

Soil fertility may vary considerably within a field. The effects of va riable soil fertility on the relationships among average crop yield re sponse, average soil test, and fertilizer applied evenly to a field ha ve not been examined. This paper develops stochastic equations to desc ribe the average yield gain on a field basis from the application of a single constant rate of fertilizer, in fields with variable soil fert ility. The equations are solved numerically for the specific case of n itrogen fertilizer on corn (Zea mays L.) in Ontario, Canada. The resul ts suggest that since the relationships among yield response, soil tes t, and applied fertilizer are non-linear, a single soil test calibrati on cannot exist for fields with different spatial variability. Soil te st calibrations obtained from sites with low variability (for example small plots) will not hold for sites with higher variability (for exam ple farm fields). Calibrations obtained from sites with low variabilit y will under-predict the optimum economic fertilizer rate for sites wi th low variability will under-predict the optimum economic fertilizer rate for sites with high variability. The results do not invalidate so il test calibration relationships per se. The challenge is to combine these calibrations with additional knowledge about the spatial distrib ution and field-scale variability of soil test values in order to maxi mize economic benefit.