Rice plants are well known for their very high demand for Si achieved by the active uptake pathways 11. The high Si accumulation in the aboveground biomass (plants) has many beneficial effects for rice plants in terms of abiotic and biotic stress mitigation 16,34. Due to the removal of aboveground biomass by harvest, rice paddy soils are prone to desilification leading to low Si availability in the paddy soils 10. Such desilification by intensive crop cultivation and harvest may result in a loss of available and amorphous Si up to more than 15 tons per hectare within five decades 35. Here we show that this desilification is limiting P availability in all paddy soils along a 2000 years chronosquence.
The Si availability in all soils of the studied chronosequence was low (most soils below 10 mg Si kg− 1) compared with other paddy soils, which contain available Si with a mean concentration of ~ 30 mg kg− 1 depending on soil pH and maximum values of > 80 mg kg− 1 36. This is in line with other studies showing pronounced Si deficiency in paddy soils 14,15. This low Si availability may even be lower during the main phase of growing season as the rate of Si uptake by rice is faster as that of dissolution of Si-containing soil minerals 37. Dissolved Si is in the form of polysilicic acid and equilibrates with monosilicic acid in soil solution depending on pH and concentration 12. The lower the Si concentration in the soil solution is, the more the silicic acid speciation shifts from polysilcic acid (with high binding affinity to soil minerals) towards monosilicic acid (with low binding affinity to soil minerals) 12. Hence, at low Si concentration in soil solution, little Si competes with P for binding to soil minerals 12, i.e., little if any P is replaced by Si from mineral binding sites, as most Si in solution is suggested to be monosilicic acid 12.
The concentrations of available P in the studied paddy soils were low. However, after Si addition, the concentration of available P increased to values above the critical value of P availability for rice cultivation of about 3.5 mg P kg− 1. Furthermore, Si addition increased the Si concentration in soil solution (Figures S2) and potentially shifted the Si speciation in soil solution toward polysilicic acid 12. This increase in Si concentration in soil solution and the potential shift toward polysilicic acid (based on higher Si concentrations in soil solution) led to P mobilization from soil minerals 38.
The biogeochemistry of paddy soils is particularly dominated by redox sensitive elements such as Fe 19. Flooding of paddy soil induces the reduction of Fe(III) and releases Fe(II) from soil minerals, reducing the binding capacity of soil oxides for P, and promoting a mobilization of P into the liquid phase. The process goes along with a reduction of Fe(III) phases and potential formation of new Fe(II) minerals, such as vivianite (Fe2 + 3(PO4)2·8H2O). It usually binds P less strongly than Fe(III) oxidic phases like ferrihydrite, goethite, lepidocrocite or hematite39.
High Si availability seems to increase the fraction of Fe(II) compared to Fe(III) (compare blue spectra Fig. 3l and x). Apart from natural soil heterogeneity at microscale 31, a higher share of Fe(II) phases may be explained by the fact that the formation of silica gel from ASi 40 favors reducing conditions 41,42, as ASi reduces the hydraulic conductivity at low matric potential and with this potentially reducing the transport of electron acceptors. Silicon also hinders the transformation of poorly crystalline minerals to more stable crystalline Fe(III) phases 43. By contrast, desilification of paddy soils can change the Fe mineralogy to more crystalline Fe(III) phases over time. As desilification of paddy soil can take place in less than 50 years of rice cultivation, the change in Fe mineralogy and related P dynamics can already happen within a few decades.
The mobilized P and Fe partly formed new particles of small sizes (tens of nanometers). The NEXAFS measurements showed that relatively larger and Fe(II)-rich particles are formed in the Si addition compared to the control treatment (Fig. 3l and x), which can be interpreted as a vivianite-like Fe(II)-phosphate phase 44. Subsequently, the mobilized P is coprecipitated in Fe(II) mineral phases as Fe(II)-P (e.g. vivianite) 45. As the Fe-P binding in Fe(II)-phosphates is weaker than in Fe(III)-phosphates 39, Fe(II)-phosphates have higher potential of re-releasing P. Consequently, the process of vivianite formation by Si addition further contributes to improved P availability. The improved P mobilization by increasing Si availability in Si-depleted soil can thus be taken into consideration as potential option to improve P use efficiency in paddy soils and therewith to reduce P fertilizer requirements.
In this study, we showed that low Si availability limits P availability in the soils of a paddy chronosequence. We attribute this finding to the fact that polysilicic acid released at high Si concentrations outcompeting P at Fe(III)-containing binding sites and that the addition of ASi, favoring reducing conditions, leads to a transformation of Fe(III) to Fe(II), so that P is either directly released into soil solution or bond less strongly to Fe(II)-containing minerals like vivianite. To increase the Si availability and the ASi amounts of paddy soils, farmers can take advantage of approaches that recycle rice straw residuals, such as rice straw burning 46 or accelerated composting 34, in order to replenish plant-available Si in soil as shown for other crops 35,47, or use different Si fertilizers 48. Optimizing Si fertilization requirements for co-beneficial needs of P fertilization in paddy soils may now warrant further attention. And finally, desilification of paddy soil eventually leads to decreased P availability and higher demand for P fertilizer in paddy soils.