SOIL SAMPLING UNDER NO-TILL BANDED PHOSPHORUS

Uncertainty exists about the best sampling procedures for no-till soils containing residual P-fertilizer bands. This study was conducted to determine the distribution of residual P-fertilizer bands and to define soil sampling procedures that minimize variability, while approximating the "true" P-soil-test value. A no-till, banded-P-fertilization field study was conducted during 1986 and 1987 on three central Great Plains soils: Keith clay loam (a fine-silty, mixed, mesic Aridic Argiustoll), Woodsen silt loam (a fine, montmorillonitic, thermic Abrubtic Argiaquoll), and Harney silt loam (a fine, montmorillonitic, mesic Typic Argiustoll). Soils were sampled laterally away from the band 1 to 2 yr after fertilization, and P was determined with NaHCO3 and acid-fluoride extractions. Bands were accurately described by the exponential decay model (mostly r2 values of 0.98 or better). The derived exponential equations were used in computer evaluation of soil-sampling procedures. When the location of P bands are known, sampling to include one in-the-band soil sample for every 20 between-the-band samples for 76-cm band spacing, and for every eight between-the-band samples for 30-cm band spacing, will result in a P soil test equal to the "true" P-soil-test mean. When the location of P bands is unknown and < 20 subsamples are taken, paired sampling consisting of a first completely random sample and a second sample 50% of the band-spacing distance from the first sample, perpendicular to the band direction, will reduce variability over completely random sampling. The greatest deviation from the "true" P soil test occurs when inadequate sampling includes rather than excludes the band, and will thus under-estimate P-fertilizer needs.