Soil phosphorus distribution and concentration from repeated starter phosphorus band application

Soil sampling to determine soil phosphorus (P) availability is complicated by the presence of residual P bands that have not been disturbed by tillage. This study investigated the soil P distribution and longevity of P bands applied in no-tillage culture on a Mollic Endoaqualf soil (fine, montmorillonitic, mesic). Starter band P treatments of 0, 5, 10, and 20 kg P ha(-1) were applied to corn (Zea mays L.) in a no-till corn-soybean [Glycine max. (L.) Merr.] rotation from 1990 through 1996. Treatments were rerandomized with each application, such that differing starter band histories were established, which served as treatments for soil and plant sampling. In 1996 the 10 and 20 kg P ha(-1) bands were marked with monofilament line in the soil during the application process. In April 1997, contiguous 2.0 cm wide, 15 cm deep soil cores (30 per row) were taken across the 1996 corn rows. Soil core Bray-1 P concentrations that were 52 to 54 mg kg(-1) greater than adjacent cores clearly indicated the presence of one and presumed three year old residual bands of 20 kg P ha(-1) treatments. Cross-sectional soil samples of these bands (1.61 cm2 samples) had P concentrations that were 35 times that of non-band affected soil, and band widths were 8 to 10 cm. When sampled by 2.0 cm diameter cores these bands were evident in only two adjoining core samples. Older greatly decayed bands were also evident by core P concentrations that were 2 to 7 mg kg(-1) greater than adjacent cores. Cross-section samples showed the bands had P concentrations that were 2 to 3 times greater than non-band affected soil. These bands were presumed to be five and seven years old. Based on the standard deviation of the 30 core sets of samples, an impractical number of samples would be required for random sampling to achieve a precision level better than 4 mg P kg(-1). Pooling all the soil across a row (a slice sample) reduced variability enough to achieve 2 mg kg(-1) precision with 95% confidence with no more than 9 slices per composite sample, and P concentrations would be increased 2 to 2.9 mg kg(-1) relative to soil without bands. In 1997 oat (Avena sativa L.) growing directly over 10 kg ha(-1) bands and non-band affected soil had similar whole plant P contents at Feekes growth stage 10. Yet the oat had incrementally greater plant P contents in response to 1997 broadcast P amounts of 0, 15, 30, and 44 kg ha(-1). Soil sampling strategies that variably weight band-affected soil relative to soil outside the band only slightly influence soil P recommendations. The choice of a sampling strategy should be based on the need for accuracy and the bias desired (inclusion or exclusion of bands). Despite the ability to measure residual soil band P to various levels of precision with sampling strategies that variably weight band affected soil in the sample, soil extractable P from bands may not well represent P actually available to plants.