Summary Most soil tests for available phosphorus (P) perform rather poorly in predicting crop response. This study was set up to compare different established soil tests in their capacity to predict crop response across contrasting types of soil. Soil samples from long‐term field experiments, the oldest >100 years old, were collected in five European countries. The total number of soil samples (n = 218), which differed in cropping and P treatment, and originated from 11 different soil types, were analysed with five tests: ammonium oxalate (Ox), ammonium lactate (AL), Olsen P, 0.01 m CaCl2 and the diffusive gradient in thin film (DGT). The first three tests denote available P quantity (Q), whereas the last two indicate P intensity (I) of the soil solution. All five tests were positively related to the crop yield data (n = 317). The Q‐tests generally outperformed I‐tests when evaluated with goodness of fit in Mitscherlich models, but critical P values of the I‐tests varied the least among different types of soil. No test was clearly superior to the others, except for the oxalate extraction, which was generally poor. The combination of Q‐ and I‐tests performed slightly better for predicting crop yield than any single soil P test. This Q + I analysis explains why recent successes with I‐tests (e.g. DGT) were found for soils with larger P sorption than for those in the present study. This systematic evaluation of soil tests using a unique compilation of established field trials provides critical soil P values that are valid across Europe. Highlights We compared soil P tests for predicting crop response across contrasting soil types. No test was clearly superior to the others except for the oxalate extraction, which was generally poor. This study suggests that intensity tests do not perform markedly better than quantity tests. The evaluation of soil P tests on this unique dataset provided critical soil P values across Europe.

The phosphate quality standards in the lowland rivers of Flanders (northern Belgium) are exceeded in over 80% of the sampling sites. The factors affecting the molybdate reactive P (MRP) in these waters were analyzed using the data of the past decade (>200 000 observations). The average MRP concentration in summer exceeds that winter by factor 3. This seasonal trend is opposite to that of the dissolved oxygen (DO) and nitrate concentrations. The negative correlations between MRP and DO is marked (r = -0.89). The MRP concentrations are geographically unrelated to erosion sensitive areas, to point-source P-emissions or to riverbed sediment P concentration. Instead, MRP concentrations significantly increase with increasing sediment P/Fe concentration ratio (p < 0.01). Laboratory static sediment-water incubations with different DO and temperature treatments confirmed suspected mechanisms: at low DO in water (<4 mg L), reductive dissolution of ferric Fe oxides was associated with mobilization of P to the water column from sediments with a molar P/Fe ratio >0.4. In contrast, no such release was found from sediments with lower P/Fe irrespective of temperature and DO treatments. This study suggests that internal loading of the legacy P in the sediments explains the MRP concentrations which are most pronounced at low DO concentrations and in regions where the P/Fe ratio in sediment is large.

No abstract

Summary The reduced use of phosphorus (P) fertilizer in fertile soil has reverted the P balance to negative in some regions. It is unclear how long current soil P stocks will ensure adequate P supply to crops. In addition, it is unknown if current soil tests for available P describe bioavailable P adequately in soil where P is becoming depleted. We set up an accelerated soil P mining test to address these questions. Perennial ryegrass (Lolium perenne, Melpetra tetra) was grown for 2 years in a greenhouse on 5‐cm‐deep soil layers of eight contrasting soils with periodic grass clipping. Each soil was split into four fertilizer treatments (i.e. no P (–P) and adequate P (+P)) and two nitrogen levels, the latter to alter the rate of P uptake. The long‐term P mining induced P‐related yield losses in seven of the 16 soil treatments. The cumulative uptake of shoot P at which yield loss started to exceed 10% (–P versus +P) varied over a small range of 37–74 mg P kg−1 soil among the soils. This critical cumulative P uptake (CCP) was related to the soil P content prior to mining measured by five soil P tests (ammonium oxalate, ammonium lactate (AL), Olsen P, 0.01 m CaCl2 and the diffusive gradient in thin film technique (DGT)); the largest R2 values were observed for P‐AL (R2 = 0.72) and P‐DGT (R2 = 0.73). However, none of the tests was diagnostic for yield loss during the depletion period. Increased N supply accelerated growth and rates of P uptake and decreased the CCP by a factor of 1.7 on average, illustrating the effect of the rate of biomass production. The CCP values obtained in the treatment with reduced N fertilizer application are likely to be the most relevant for the field and suggest that current stocks allow adequate P supply for arable crops for 3–8 years under zero P application (0–23 cm) in soils similar to those tested. The lack of a successful diagnosis for P deficiency during this depletion experiment calls for further calibration of soil tests for available P in the field. Highlights The availability of legacy P in well‐fertilized soil was evaluated with a P mining pot trial 10% loss of crop growth occurred when soil P was depleted by 37–74 mg P kg−1 soil Accelerated plant growth with increased N supply decreased total P uptake beyond which P deficiency occurs In a depletion scenario, current soil P tests are not diagnostic but they can be used for prediction

This study was set up to identify the role of the phosphorus (P) desorption rate in P diffusion and in P bioavailability in soil. The P desorption kinetics were measured with a zero‐sink method in soil suspensions (0–77 days) for a set of soils that either had or had not been mined for P in a glasshouse study. The desorption kinetics was fitted by a serial two‐pool model, discriminating a fast desorbing P pool (Q1) with desorption half‐lives of 3–8 days, and a slowly desorbing P pool (Q2), which replenishes the fast P pool with 100‐fold larger half‐lives than the fast pool. Phosphate desorption was smaller and slower after soil P mining compared to that in the original soil samples and mining reduced the Q1/Q2 ratio. This kinetic model was embedded in a 1D planar diffusion model predicting that the diffusive flux of P to a zero sink in 5 days varies by a factor of 1.4 among the observed Q1 desorption rate constants, keeping other parameters constant, and that the reduced Q1/Q2 ratio upon P mining sharply reduces the diffusible P in soil. The P uptake model of Barber‐Cushman was extended with P desorption kinetics and was successfully calibrated to the P uptake data of the glasshouse P mining study. The model correctly predicted that reduced nitrogen (N) fertilization enhances the soil P‐use efficiency because of lower critical P demand rates at slower growth. Finally, that new model predicted that maize requires >3‐fold more available P in soil than wheat because of a higher P demand rate per unit root area of maize than that of wheat. This confirms a similar factor difference in critical soil P concentrations observed in P‐response trials in Belgium between 1973 and 2018. This study shows that the P desorption rate limits P bioavailability for fast growing plants with a small effective root area, especially under negative soil P balances that slow down the desorption rate of P in soil. Highlights The diffusion coefficient of P in soil is reduced by soil P mining Faster growing plants require more available P in soil because they rely on high diffusive P fluxes P desorption rate can limit the P bioavailability