The role of soil organic matter on the long-term availability of phosphorus in agricultural soils

Decades of excessive use of phosphorus (P) fertilisers in agriculture resulted in large soil P reserves in arable soils. This historical over-fertilisation contrasts with the impending scarcity of natural mineral P resources and with the adverse environmental effects due to losses of P to the natural environment; for these reasons, the reservoirs of soil P are often denoted as the legacy P. To reduce these environmental effects, fertilisation policies in some regions advocate a negative annual P balance on agricultural soils, thereby mining the legacy P from the soil. Critical in this undertaking is that a high bioavailability of P to crops is ensured. However, it is well known that historically added P in soil has a reduced availability compared to that of a within-season P application due to ageing processes ("fixation") of P in soil. To guarantee adequate soil fertility during P mining, an accurate evaluation of the long-term availability of the legacy P in soils is needed. This thesis is devoted to this evaluation, with a specific objective to identify to what extent soil organic carbon (SOC) or the source of P application (mineral vs organic) affects the long-term availability of these P stocks. The rationale is twofold: (1) from a soil chemical point of view, it is expected that the hysteresis encountered during mining is less pronounced in soils with a high SOC content due to the competition of organic matter with phosphate (PO4) for sorption sites, which reduces ageing of P; (2) organic P (Po), which is generally less prone to ageing than inorganic PO4, might contribute significantly to plant P availability, especially under conditions of reduced inorganic P (Pi) availability such as during a P mining scenario. Both aspects were tackled in this thesis.Assessment of the potential contribution of organic P to P availability in soil requires the quantification of net Po mineralisation rates. This detection of Po mineralisation in soil is, however, complex because it is often hampered by the fast and strong immobilisation of PO4 that is released to the soil solution. Radioisotope dilution methods have been used to overcome this issue, but remain challenged by adequate accounting for the physicochemical reactions of PO4 in soil. In this thesis, an isotope (32P) dilution approach with compartmental modelling was tested that does not require soil sterilisation. Six soils with contrasting P availability and SOC content were incubated with 32P and its dilution in five biologically or physically relevant soil P pools was monitored over time. Compartmental modelling allowed testing of several model set-ups, varying in the number of included P pools and inter-pool transformations, and evaluating them for their potential to accurately describe the dynamics of P in soil. With the most suitable model, gross and net Po mineralisation rates were quantified by fitting to the data from the incubation experiment. Biological fluxes, i.e. immobilisation into and mineralisation of microbial and organic P, were consistently smaller than physicochemical fluxes. Net organic P mineralisation rates ranged between -0.26 and +0.22 mg P kg-1 day-1 depending on the soil and increased upon amendment with maize. However, the approach met with large uncertainties, particularly on the biological reactions, and none of the observed net rates were significantly different from zero. As an alternative, a new isotope method was proposed, using - for the very first time in soil - PO4 enriched in the stable oxygen-18 isotope (18Op) in addition to the conventionally used radioactive 32P that requires soil sterilisation to identify the abiotic reactions. The fate of this 18Op depends on the type of reaction in which PO4 is involved and can, therefore, allow to disentangle biological from physicochemical isotope dilution reactions. An incubation experiment was set up with an arable soil that was either sterilised or not, and subsequently spiked with 32P and 18Op at a total P dose of ~ 5 mg PO4-P kg dry soil-1. The dilution of both isotopes by P transformation reactions was monitored in the resin extractable P pool over the course of 50 days. The gradual isotope dilution of 18Op was equal to that of 32P in the sterilised soil, confirming methodological accuracy. The biological reactions enhanced the dilution of 18Op compared to that of 32P by exchange of oxygen with water; that difference widened when microbial activity was stimulated with glutamate. By compartmental modelling, this distinct behaviour was used to quantify soil P transformation rates in the non-sterilised soils without using sterile soil data. A gross immobilisation and mineralisation rate of ~0.9 mg P kg-1 day-1 were estimated, these were a factor 5 lower than corresponding physicochemical rates and were borderline detectable in contrast with the single isotope method with sterilisation that yielded spurious data. The resulting net organic P mineralisation rate was low and undetectable (not different from zero). Despite the potential in double labelling, statistical uncertainties on each of the reaction rates remained high. To conclude, detection of soil Po mineralisation in agricultural soils remains complex, even with the two alternative methods. Nevertheless, the obtained results were deemed realistic and suggested that net Po mineralisation in soils is smaller than estimated by earlier methods and rather tends towards zero.Second, the effect of SOC on ageing processes was assessed, the premise being that OC may limit irreversible P fixation in soil by blocking P sorption sites on sesquioxides. This was first addressed in a chemical mining experiment, in which a large collection of soils, composed from experimentally amended soils (mineral vs organic field treatments and incubation trials) and soils with contrasting properties, was subjected to prolonged P desorption using anion exchange membranes as a zero P sink. Cumulative P desorption data were fitted with a two-pool kinetic desorption model, yielding estimates for a fast (labile) and total desorbable P pool. On average, 42% of P associated with poorly crystalline iron (Fe) and aluminium (Al) (oxy)hydroxides (oxalate-extractable P; Pox) was desorbable and 25% of that fraction (i.e. 11% of Pox) was labile. That labile P pool matched well with the 24h isotopically exchangeable P in these soils (R2=0.74). Both the fast and total desorbable fraction of Pox were larger at higher degrees of phosphorus saturation (DPS). In soils with a low DPS (< 0.30), the labile fraction of Pox increased as the ratio of OC to Feox and Alox increased (R2 = 0.70; p<0.001), but soils with a higher DPS did not exhibit that trend. As a mechanism, it was hypothesised that SOC molecules can effectively prevent diffusion of P into micropores of soil reactive minerals by spreading out over the sorption surface, thereby reducing P ageing. In more P saturated soils, such as are prevalent in Western Europe, the lack of available sorption sites likely hinders spreading out of the SOC by competitive and electrostatic effects of sorbed P. In these soils, the DPS remained the main soil characteristic explaining the variability in the available P fractions. Similar trends were observed in an accelerated biological mining experiment using ryegrass to mine P from largely the same set of soils. In this 15-months long experiment, the long-term availability of P to plants was evaluated by determining adequate and total P uptake and correlating these to selected soil properties, with due attention to SOC. The fractions of plant-available Pox indeed increased with increasing labile P fractions determined by chemical mining and at increasing ratios of P and OC to Feox and Alox. This supports the hypothesis that increased SOC contents reduce ageing of P by preventing its diffusion into micropores. A comparison of the results from this biological mining with available soil P pools determined in the (sterile) P desorption experiment suggested an insignificant contribution of organic P to plant P supply.Taken together, the findings of this work suggest that legacy P from well-fertilised agricultural soils could act as a sufficient P source for plants for 12-175 years, depending on the soil. This long-term availability is positively affected by higher soil organic matter contents, i.e. the legacy P will have a higher fertiliser efficiency in soils with a high content of SOC per unit of Fe and Al in soil, as long as the soil is not yet too saturated with P. A significant contribution of organic P mineralisation to plant P availability could not be confirmed in the biological mining experiment, which was in line with the small net mineralisation rates estimated with both compartmental models tested in this thesis. Translated to the field, our results suggest that soil management practices stimulating SOC sequestration and application of organic instead of mineral fertilisers during the period of build-up could be beneficial in terms of availability, provided that the soil does not become too saturated with P.

Publication year:2021

Accessibility:Open