Increasing phosphorus fertilizer value of recycled iron phosphates by prolonged flooding and organic matter addition

Iron(Fe)minerals are commonly used to remove phosphorus(P)from waste streams,producing P-loaded Fe(Ⅲ)oxides or Fe(Ⅱ)phosphate minerals(e.g.,vivianite).These minerals may be used as fertilizers to enhance P circularity if solubilized in soil.Here,we tested the P fertilizer value of recycled Fe phosphates(FePs)in a pot trial and in an incubation experiment,hypothesizing that P release from FePs is possible under Fe(Ⅲ)-reducing conditions.First,a pot trial was set up with rice(Oryza sativa)in all combinations of soil flooding or not,three P-deficient soils(acid,neutral,and calcareous),and six FePs(three Fe(Ⅲ)Ps and three Fe(Ⅱ)Ps)referenced to triple superphosphate(TSP)or zero amendments.Shoot P uptake responded to TSP application in all treatments but only marginally to FePs.The redox potential did not decrease to-200 mV by flooding for a brief period(13 d)during the pot trial.A longer incubation experiment(60 d)was performed,including a treatment of glutamate addition to stimulate reductive conditions,and P availability was assessed with CaCl_(2)extraction of soils.Glutamate addition and/or longer incubation lowered soil redox potential to<-100 mV.On the longer term,Fe(Ⅲ)minerals released P,and adequate P was reached in the calcareous soil and in the neutral soil amended with Fe(Ⅲ)P-sludge.It can be concluded that prolonged soil flooding and organic matter addition can enhance the P fertilizer efficiency of FePs.Additionally,application of FeP in powder form may enhance P availability.

Prediction of exchangeable potassium in soil through mid-infrared spectroscopy and deep learning:From prediction to explainability

The ability to characterize rapidly and repeatedly exchangeable potassium(Kex)content in the soil is essential for optimizing remediation of radiocaesium contamination in agriculture.In this paper,we show how this can be now achieved using a Convolutional Neural Network(CNN)model trained on a large Mid-Infrared(MIR)soil spectral library(40,000 samples with Kex determined with 1 M NH4OAc,pH 7),compiled by the National Soil Survey Center of the United States Department of Agriculture.Using Partial Least Squares Regression as a base-line,we found that our implemented CNN leads to a significantly higher prediction performance of Kex when a large amount of data is available(10000),increasing the coefficient of determination from 0.64 to 0.79,and reducing the Mean Absolute Percentage Error from 135%to 31%.Furthermore,in order to provide end-users with required interpretive keys,we implemented the GradientShap algorithm to identify the spectral regions considered important by the model for predicting Kex.Used in the context of the implemented CNN on various Soil Taxonomy Orders,it allowed(i)to relate the important spectral features to domain knowledge and(ii)to demonstrate that including all Soil Taxonomy Orders in CNN-based modeling is beneficial as spectral features learned can be reused across different,sometimes underrepresented orders.