Chlordecone (CLD), was widely applied in banana fields in the French West Indies from 1972 to 1993. The WISORCH model was constructed to assess soil contamination by CLD and estimated that it lasts from 100 to 600 years, depending on leaching intensity and assuming no degradation. However, recent studies demonstrated that CLD is degraded in the environment, hence questioning the reliability of previous estimations. This paper shows how to improve the model and provides insights into the long-term dissipation of CLD. In-situ observations were made in nearly 2545 plots between 2001 and 2020, and 17 plots were sampled at two dates. Results of soil analyses showed an unexpected 4-fold decrease in CLD concentrations in the soil, in contrast to simulations made using the first version of WISORCH at the time. Neither erosion, nor CLD leaching explained these discrepancies. In a top-down modeling approach, these new observations of CLD concentrations led us to implement a new dissipation process in the WISORCH model that corresponds to a DT50 dissipation half-life of 5 years. The new version of the improved model allowed us to update the prediction of the persistence of soil pollution, with soil decontamination estimated for the 2070s. This development calls for re-evaluation of soil pollution status. Further validation of the new version of WISORCH is needed so it can contribute to crop management on contaminated soil.

The In Situ Chemical Reduction (ISCR) process was tested in a nitisol in a French Caribbean banana plantation using five different soil amendments. The addition of 2.8% or 4.0% of Zero Valent Iron (ZVI; dw/dw, 2 different trial plots) in the 0–40-cm soil layer lowered the initial chlordecone (CLD) concentration by up to 74% or 69% in 37 days or 94 days, with 75% of the decrease achieved after only 21 or 24 days of treatment depending on the trial plot. The addition of commercially available Daramend® was also tested by applying the 6% dose (dw/dw) recommended by the manufacturer and using either the regular alfalfa-based product or a bagasse-based product specifically formulated for the study. Both significantly lowered CLD concentrations, but to a lesser extent than with the ZVI-only amendment. A bagasse-ZVI mixture prepared on site produced results slightly better than the two Daramend®. The percentage decreases in CLD concentrations were correlated with the negative redox potentials achieved. In all the trial plots, dechlorinated transformation products appeared in the soil and soil water as the CLD concentrations decreased, with H atoms replacing up to 4 and 7 of the 10 Cl atoms, respectively. None of these degradation products appeared to accumulate in the soil or soil water during the treatment. Instead, the reverse occurred, with an overall downward trend in their concentrations over time. The effects of ISCR treatment on agronomic and human health–related parameters were measured in three different crops. The radishes produced with some treatments were visually of lower quality or smaller in size than those grown in the control plots. Lower yields were observed for the cucumbers and sweet potatoes grown after applying the bagasse-based amendments. Mortality among cucumber seedlings was observed after treatment with ZVI only. Simple operational solutions should suffice to remedy these negative agronomic effects. As regards human health–related effects, the CLD concentrations in radishes grown with three of the amendments were significantly lower than in the two control plots and well below the maximum residue level (MRL), which was substantially exceeded in the radishes grown on untreated soil. For cucumbers, the treatments with regular Daramend® and with a local bagasse-ZVI mixture produced fruits with CLD below the MRL and also below the concentrations in one of the two control plots. As for the sweet potatoes, adding a bagasse-ZVI mixture had a significant positive effect by decreasing contamination below the levels in the two control plots and below the MRL.

In the French West Indies, the chlordecone (organochloride pesticide) pollution is now diffuse becoming new contamination source for crops and environment (water, trophic chain). Decontamination by bioremediation and chemical degradation are still under development but the physical limitations of these approaches are generally not taken into account. These physical limitations are related to the poor physical accessibility to the pesticides in soils because of the peculiar structural properties of the contaminated clays (pore volume, transport properties, permeability, and diffusion). Some volcanic soils (andosols), which represent the half of the contaminated soils in Martinique, contain nanoclay (allophane) with a unique structure and porous properties. Andosols are characterized by pore size distribution in the mesoporous range, a high specific surface area, a large pore volume, and a fractal structure. Our hypothesis is that the clay microstructure characteristics are crucial physico-chemical factors strongly limiting the remediation of the pesticide. Our results show that allophane microstructure (small pore size, hierarchical microstructure, and tortuosity) favors accumulation of chlordecone, in andosols. Moreover, the clay microporosity limits the accessibility of microorganisms and chemical species able to decontaminate because of poor transport properties (permeability and diffusion). We model the transport properties by two approaches: (1) we use a numerical model to simulate the structure of allophane aggregates. The algorithm is based on a cluster–cluster aggregation model. From the simulated data, we derived the pore volume, specific surface area, tortuosity, permeability, and diffusion. We show that transport properties strongly decrease because of the presence of allophane. (2) The fractal approach. We characterize the fractal features (size of the fractal aggregate, fractal dimension, tortuosity inside allophane aggregates) and we calculate that transport properties decrease of several order ranges inside the clay aggregates. These poor transport properties are important parameters to explain the poor accessibility to pollutants in volcanic soils and should be taken into account by future decontamination process. We conclude that for andosols, this inaccessibility could render inefficient some of the methods proposed in the literature.