Skip to Main content Skip to Navigation
Theses

Elaboration d'un modèle de prédiction de la phytodisponibilité du cadmium dans les sols agricoles : application à la contamination cadmiée du blé dur

Abstract : Cadmium (Cd) is a trace element found in agricultural soils which can contaminate the food chain by being taken up by plants and accumulated in consumed plant products. The bioavailability of Cd is a concept at the centre of the risk assessment of Cd transfer from soil to plants. Plants, taking up essentially the free form of Cd (Cd2+) in the soil solution, bioavailability is a function of two main processes, the Cd speciation in solution (the different chemical forms taken by the Cd in solution) and the partitioning of Cd2+ between the solid phase and the solution. The main objective of this work was to develop simple predictive operational models of Cd2+ concentrations in agricultural soil solution, by modelling either Cd speciation in soil solution or Cd2+ solid-solution partitioning, the latter to further inform the ability of the phase to replenish the soil solution during root absorption. We also looked for models for the prediction of Cd2+ levels found in durum wheat grains. We have developed two modelling approaches. The first, statistical, allows to produce simple models for operational purposes. The second, geochemical, allows to understand the dominant mechanisms and thus to judge the coherence of simple statistical models to represent complex physicochemical processes. These two approaches have consistently shown that for poorly contaminated agricultural soils, the exchanged Cd between the solid phase and the solution is likely to be weakly sorbed and can be estimated by the Cd extracted by 1M NH4NO3 minus a fraction attached to amorphous manganese oxides. Modelling also showed the importance of pH and Ca content in solution as variables controlling the solubility of Cd2+, probably because of their role in controlling the sorption site availability and Cd complexation. The best-performing statistical model for predicting Cd content in durum wheat also picks up these variables, suggesting that simple statistical models for speciation in solution and soil-solution partitioning of Cd are relevant models to estimate bioavailability and that they can be used to classify soils according to the risks of the transfer of soil Cd to a crop. Compared to the evolution of the regulatory context, the statistical model predictive of the Cd content in the grain shown by cross validation that it could discern relatively reliably (88%) the differences of 0.05 mg Cd.kg-1 of grain and that its reliability would be less (65%) for differences of 0.025 mg Cd.kg-1.
Complete list of metadata

Cited literature [373 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-01895809
Contributor : Abes Star :  Contact
Submitted on : Monday, October 15, 2018 - 3:10:30 PM
Last modification on : Friday, January 29, 2021 - 11:30:03 AM
Long-term archiving on: : Wednesday, January 16, 2019 - 3:11:21 PM

File

VIALA_YOANN_2018_CORR.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-01895809, version 1

Collections

Citation

Yoann Viala. Elaboration d'un modèle de prédiction de la phytodisponibilité du cadmium dans les sols agricoles : application à la contamination cadmiée du blé dur. Biodiversité et Ecologie. Université de Bordeaux, 2018. Français. ⟨NNT : 2018BORD0091⟩. ⟨tel-01895809⟩

Share

Metrics

Record views

289

Files downloads

414