Wheat yield prediction by zero sink and equilibrium-type soil phosphorus tests

Diffusive gradients in thin films (DGT) measurements have been shown to outperform other phosphorus (P) tests in soils with strong P sorption, but this has not been confirmed for moderately weathered European soils. We compared the performance of DGT in predicting wheat grain yield in Swedish long-term fertility experiments with those of standard intensity (water-extractable P (P-H2O)) and quantity (ammonium lactate-extractable P (P-AL)) tests. A Mitscherlich-type model was used to fit wheat yield response to P application rates (0, 15, 30 or 35, and 45 kg P ha-1 year-1) in each individual trial replicate to estimate the maximum yield. For trials with clear plateau-type yield responses and the goodness of fit (R2) > 0.75, relative yields (RYs) were calculated for each P treatment and plotted against the soil P test results (n = 143). The goodness of the Mitscherlich-type fits decreased in the following order: DGT-measured P (P-DGT) (R2 = 0.35) > P-H2O (R2 = 0.18) > P-AL (R2 = 0.13). When excluding soils with P-AL:P-DGT ≥ 0.1 L g-1, R2 was considerably improved to 0.55 for P-AL, 0.46 for P-H2O, and 0.65 for P-DGT (n = 61). At 95% of maximum yield, the upper limit of P deficiency for P-DGT was 44.8 (the soils with P-AL:P-DGT < 0.1 L g-1) and 61.9 μg L-1 (all soils), falling within the range reported for other European and Australian soils (6.0–142 μg L-1). We show that in the investigated Swedish soils, DGT performed better than the quantity and intensity tests, which is attributed to its ability to capture P diffusion and resupply from the soil solid phase, similar to plant roots in the rhizosphere.