Determining Economic Optimum Soil Sampling Density for Potassium Fertilizer Management in Soybean: A Case Study in the U.S. Mid-South

Abstract

Determining the number of samples to collect in a field to develop soil-test K (STK) maps that are sufficiently accurate for profit-maximizing fertilizer rate prescription maps is complex. The decision also hinges on the application method—variable rate or uniform rate (VRT vs. URT). Using a 400 m2 fishnet grid on a 26.3-ha irrigated soybean field, the authors compared sampling densities ranging from 5 to 60 samples or 5.3 ha/sample to 0.40 ha/sample. Subsequently, the authors simulated yields based on STK maps generated with that range of samples taken to generate i) associated profit-maximizing fertilizer-K rates (K*) that varied by grid with VRT, or ii) a single fertilizer rate based on field-average STK with URT, to compare revenue less fertilizer cost (NR) across VRT, URT, and sampling strategy. With more information, NR increased at a diminishing rate as crop needs could be better matched to fertilizer needs with greater detail in STK maps with VRT. Also, fertilizer use with URT was higher than VRT given the field-specific distribution of STK. Regardless of the sampling strategy, NR was higher for VRT than URT, however, that benefit was smaller than the upcharges for VRT equipment. Marginal benefits from added soil sampling were smaller than their marginal cost leading to an optimal least-cost, 5-sample strategy and URT. Changing one of the 5 sampling locations, however, revealed unreliable field average STK estimates. Since soil samples inform about several macronutrients, splitting soil sampling charges across K and P profitably justified sampling near every 1.5 ha with URT.