Prediction of nano, fine, and medium colloidal phosphorus in agricultural soils with machine learning.

Soil colloids have been shown to play a critical role in soil phosphorus (P) mobility and transport. However, identifying the potential mechanisms behind colloidal P (Pcoll) release and the key influencing factors remains a blind spot. Herein, a machine learning approach (random forest (RF) coupled with partial dependence plot analyses) was applied to determine the effects of different soil physicochemical parameters on Pcoll content in three colloidal subfractions (i.e., nano- (NC): 1-20 nm, fine- (FC): 20-220 nm and medium-sized colloids (MC): 220-450 nm) based on a regional dataset of 12 farmlands in Zhejiang Province, China. RF successfully predicted Pcoll content (R2 = 0.98). Results showed that colloidal- organic carbon (OCcoll) and minerals were the major determinants of total Pcoll content (1-450 nm); their critical values for increasing Pcoll release were 87.0 mg L-1 for OCcoll, 11.0 mg L-1 for iron (Fecoll) or aluminium (Alcoll), 2.6 mg L-1 for calcium (Cacoll), 9.0 mg L-1 for magnesium (Mgcoll), 2.5 mg L-1 for silicon (Sicoll), and 1.4 mg L-1 for manganese (Mncoll). Among three colloidal subfractions, the major factors determining Pcoll were soil Olsen-P (POlsen; 125.0 mg kg-1), Cacoll (2.5 mg L-1), and colloidal P saturation (21.0%) in NC; Mncoll (1.5 mg L-1), Mgcoll (6.8 mg L-1), and POlsen (135.0 mg kg-1) in FC; while Mncoll (1.5 mg L-1), Alcoll (2.5 mg L-1), and Fecoll (3.8 mg L-1) in MC, respectively. OCcoll had a considerable effect in the three fractions, with critical values of 80.0 mg L-1 in NC or FC, and 50.0 mg L-1 in MC. Our study concluded that the information gleaned using the RF model can be used as crucial evidence to identify the key determinants of different size fractionated Pcoll contents. However, we still need to discover one or more easy-to-measure parameters that can help us better predict Pcoll.

Nano and fine colloids suspended in the soil solution regulate phosphorus desorption and lability in organic fertiliser-amended soils.

Mobile colloids impact phosphorus (P) binding and transport in agroecosystems. However, their relationship to P-lability and their relative importance to P-bioavailability is unclear. In soils amended with organic fertilisers, we investigated the effects of nano (NC; 1-20 nm), fine (FC; 20-220 nm), and medium (MC; 220-450 nm) colloids suspended in soil solution on soil P-desorption and lability. The underlying hypothesis is that mobile colloids of different sizes, i.e., NC, FC, and MC, may contribute differently to P-lability in soils enriched with organic fertiliser. NC- and FC-bound Pcoll were positively correlated with P-lability parameters from diffusive gradient in thin films (DGTA-labile P concentration, r ≥ 0.88; and DGTA-effective P concentration, r ≥ 0.87). The corresponding relations with MC-bound Pcoll are weaker (r values of 0.50 and 0.51). NC- and FC-bound Pcoll were also strongly correlated with soil P-resupply (r ≥ 0.64) and desorption (r ≥ 0.79) parameters during DGTA deployment, and the mobility of these colloids was corroborated by electron microscopy of DGTA gels. MC-bound Pcoll was negatively correlated with the solid-to-solution distribution coefficient (r = -0.42), indicating this fraction is unlikely to be the source of P-release from the solid phase after P-depletion from the soil solution. We conclude that NC and FC mainly contribute to regulating soil desorbable-P supply to the soil solution in the DGTA depletion zone (in vitro proxy for plant rhizosphere), and consequently may act as critical conditioners of P-bioavailability, whereas MC tends to form complexes that lead to P-occlusion rather than lability.