Environ Sci Pollut Res DOI 10.1007/s11356-016-7582-4

4TH INTERNATIONAL SYMPOSIUM ON ENVIRONMENTAL BIOTECHNOLOGY AND ENGINEERING-2014

Remediation by chemical reduction in laboratory mesocosms of three chlordecone-contaminated tropical soils Christophe Mouvet 1 & Marie-Christine Dictor 1 & Sébastien Bristeau 2 & Dominique Breeze 2 & Anne Mercier 1

Received: 17 March 2016 / Accepted: 2 September 2016 # Springer-Verlag Berlin Heidelberg 2016

Abstract Chlordecone (CLD), a highly persistent organochlorine pesticide commonly encountered in French West Indies (FWI) agricultural soils, represents a major source of contamination of FWI ecosystems. The potential of chemical reduction for remediation of CLD-contaminated soil has been investigated in laboratory pilot-scale 80 kg mesocosms for andosol, ferralsol, and nitisol from FWI banana plantations. Six cycles consisting of a 3-week reducing phase followed by a 1-week oxidizing phase were applied, with 2 % (dw/dw) Daramend® (organic plant matter fortified with zero valent iron) added at the start of each cycle. Complementary amendments of zero valent iron and zinc (total of 3 % dw/dw) were added at the start of the first three cycles. After the 6-month treatment, the CLD soil concentration was lowered by 74 % in nitisol, 71 % in ferralsol, and 22 % in andosol. Eleven CLDdechlorinated transformation products, from mono- to pentadechlorinated, were identified. None of them accumulated over the duration of the experiment. Six of the seven ecotoxicological tests applied showed no difference between the control and treated soils. The treatment applied in this study

Responsible editor: Philippe Garrigues Electronic supplementary material The online version of this article (doi:10.1007/s11356-016-7582-4) contains supplementary material, which is available to authorized users. * Christophe Mouvet [email protected]

1

BRGM – Water, Environment and Ecotechnologies Division, 3 Av. Claude Guillemin, 45060 Orléans, Cedex 2, France

2

BRGM – Laboratory Division, 3 Av. Claude Guillemin, 45060 Orléans, Cedex 2, France

may offer a means to remediate CLD-contaminated soils, especially nitisol and ferralsol. Keywords Chlordecone . Chemical reduction . Soil remediation . Daramend® . Zero-valent iron

Introduction The extensive use of the insecticide chlordecone (CLD; also known as Kepone®; C10Cl10O; CAS number: 143–50-0 and CAS name: 1,1a,3,3a,4,5,5,5a,5b,6-decachlorooctahydro1,3,4-metheno-2H-cyclobuta[cd]pentalen-2-one) to fight the banana root borer Cosmopolites sordidus in the French West Indies (FWI) islands during the periods 1972–1978 and 1981– 1993 has resulted in contamination of soils (Cabidoche et al. 2009), surface and ground waters (Bocquené and Franco 2005; Gourcy et al. 2009), livestock (Jondreville et al. 2014), freshwater fish and crustaceans (Monti and Coat 2007), and halieutic fish and seafood resources (Bertrand et al. 2009; Coat et al. 2006, 2011). The arable land area at high risk of CLD contamination in the FWI represents hundreds of km2, about 15–19 % of the total arable land area, and often covers the recharge areas of ground and surface water abstraction wells (Colombano et al. 2009). Without the implementation of remediation, contamination is expected to last for decades to centuries according to soil type (Cabidoche et al. 2009). In 2002, the CLD contamination of root vegetables led regulatory authorities in FWI to take precautionary ordinances, which restricted the marketing of food products contingent (Achard et al. 2007). Statutory limitations on human consumption were issued for vegetables in 2003 (Joly 2010) and poultry and other animal food in 2005 (Journal official de la République Française 2005). Recently, bans on

Environ Sci Pollut Res

consumption and commercialization of fish and seafood have also been issued (Préfecture de la Région Guadeloupe 2010). An increased occurrence of prostate cancer and preterm birth as well as negative impact on the cognitive, visual and motor development of infants have been demonstrated (Multigner et al. 2010; Dallaire et al. 2012; Kadhel et al. 2014). It is therefore clear that CLD contamination of soils in the FWI is a major environmental issue that must urgently be addressed. Work on this issue will also have beneficial outcomes in other parts of the world. Indeed, CLD has also been used in Cameroon, Ivory Coast, Equator, Honduras, Nicaragua, Panama (Fritz 2009; OPECST 2009; Joly 2010) and in the USA until 1976 as well as in Asia (UNEP/POPS/ POPRC 2007). In terms of remedial action, the Bdig-and-dump^ approach for CLD-impacted soils is not practical in the FWI due to the magnitude of the problem, access issues, and resource constraints. Conventional bioremediation technologies and phytoremediation are not appropriate either (Colombano et al. 2009). Sequestration of CLD achieved by adding compost to andosol and nitisol soils resulted in a decreased uptake by vegetables and reduced the leaching by water (Woignier et al. 2012, 2013). However, remobilization of CLD could take place as the compost is being mineralized; hence, longlasting sequestration was not validated. As for In Situ Chemical Reduction (ISCR), it has been identified as a route likely to enable field application for remediation of soils contaminated by CLD, favoring its transformation (Clostre et al. 2010). Sequestration and transformation might be complementary, depending on the objectives. ISCR as defined originally (United States Patent 2000) is based on (multiple) treatment cycles of alternating strongly reducing conditions (brought about by the addition of Daramend® amendments and water to reach water content close to the soil water holding capacity) with oxidizing conditions (induced via tilling). To facilitate ISCR, the Daramend® amendments combine controlled-release carbon with a reduced metal—such as zero valent iron (ZVI) or zinc (ZVZ)—to stimulate the degradation of persistent organic compounds without accumulation of catabolic intermediates (Seech et al. 2008). ISCR has been proven effective in remediation of soils contaminated by PCBs (Abbey et al. 2003), pesticides (Phillips et al. 2004, 2005, 2006; Kim et al. 2010) and explosives (Elgh Dalgren et al. 2009; Zhuang et al. 2014). The contaminants and the soils studied above are quite different from the FWI CLD situation. Indeed, the dishomocubane skeleton of CLD is very different from that of the aforementioned molecules. Additionally, the stability of CLD (UNEP/POPS/POPRC 2005) and its strong sorption to soils may reduce the yield of remediation by ISCR. As for the soils studied here, some of their characteristics (andosol: very high organic matter content and specific clays with extremely high porosity trapping organic carbon and contaminants

(Woignier et al. 2012; Chevallier et al. 2010)) are very different from those of the soils for which ISCR has been validated. The objective of this study was therefore to evaluate the performance of chemical reduction applied to the three major CLD-contaminated soil types of the FWI. The study was conducted in mesocosms under controlled laboratory conditions, as a first step before field trials. In addition to the monitoring of major physico-chemical parameters and CLD and transformation products, ecotoxicity tests were also conducted on soils or soil extracts, using standardized protocols to assess the overall impact of the remediation treatment on various target organisms (Pandard et al. 2006; Römbke et al. 2009; Moser et al. 2011).

Materials and methods Field sites and soil sampling The three major soil types on which bananas are grown (andosol, AND; ferralsol, FRL; nitisol, NIT) were collected in Guadeloupe following the sampling strategy developed to assess the spatial variability of CLD concentration in banana fields (Achard et al. 2003; Clostre et al. 2014). In brief, 25, 8kg samples of the top 30 cm taken from a ca. 1 ha area were combined and homogenized to yield a single, composite sample of 400–550 kg (according to soil type) fresh weight soil. The total mass of each soil type was manually homogenized for removing coarse stones and vegetable residues. Subsamples were subsequently taken by quartering for the determination of soil properties and for filling-up the mesocosms used in the experimental remediation process. Determinations of major physico-chemical parameters (Table 1) were conducted by an accredited laboratory (SAS, Ardon, France) using standardized procedures previously reported (Mercier et al. 2013). Procedure for soil treatment For each of the three soil types, two individual mesocosms (one control, one treated by chemical reduction) were made with 80 kg of homogenized sub-samples. The six mesocosms were kept at 25 ± 2 °C (covering the range of soil temperature at 10–30 cm depth in the Martinique Island; Baran and Barras 2008). The other treatment conditions were selected by expert advice from Dr. Jim Muller (President of Provectus Environmental Products™, formerly Director of Remedial Solutions & Strategies, Adventus Americas) and following preliminary short-term trials in 2-L size microcosms. The treatment consisted of 6 cycles (SM1 Fig. S1) with unit duration of 4 weeks: 3 weeks of a reducing phase followed by 1 week of an oxidizing phase. Each cycle was initiated via the addition of 2 % (dw/dw) Daramend® (hereafter: Bthe

Environ Sci Pollut Res Table 1 Physico-chemical characteristics of the three Guadeloupean soils studied (mean ± standard deviation, n = 3; from Mercier et al. 2013)

Parameters

Soil type Andosol

Ferralsol

Nitisol

Coarse sand (%)

9.9 ± 1.2

4.3 ± 0.7

Fine sand (%)

6.3 ± 1.0

2.7 ± 0.5

6.9 ± 0.4

Coarse silt (%)

27.0 ± 1.9

13.7 ± 2.3

17.4 ± 0.9

Fine silt (%) Clay (%)

19.7 ± 0.7 23.6 ± 0.5

16.4 ± 0.3 59.1 ± 2.3

24.9 ± 0.7 37.8 ± 1.3

pH water Water content (%)

5.3 ± 0.0 60.3 ± 0.1

5.6 ± 0.1 30.1 ± 1.0

6.1 ± 0.0 27.0 ± 0.3

Organic matter (%)

13.4 ± 0.2 37.5 ± 2.2

3.9 ± 0.1 17.2 ± 0.4

3.7 ± 0.0 26.9 ± 0.2

0.7 ± 0.0 7.8 ± 0.1

0.2 ± 0.0 2.1 ± 0.1

0.2 ± 0.0 2.3 ± 0.1

C/N Total Fe (%)

11.2 ± 0.2 7.6 ± 0.3

10.5 ± 0.7 11.1 ± 0.3

11.3 ± 0.4 9.2 ± 0.3

Total Al (%) Chlordecone concentration (mg kg−1)

9.1 ± 0.2 14.3 ± 1.1

11.5 ± 0.3 2.6 ± 0.1

10.4 ± 0.2 1.2 ± 0.2

Cation exchange capacity (meq 100 g−1) Total N (%) Organic C (%)

amendment,^ 55 % ground leguminous plant and 45 % iron, dw/dw) at the beginning of the reducing phase. The soils were brought to a target percent (90 % for cycles 1 and 2) of their water holding capacity (WHC) by addition of tap water using a 5-L sprayer. At the onset of the 1st, 2nd and 3rd cycle, 2 % (dw/dw) zero valent iron (ZVI), 0.03 % zero valent zinc, and 1 % ZVI were added along with the amendment, respectively. The additions of zero valent metals complementary to the amendment aimed at maximizing the decrease in redox potential. For the 3rd cycle, 120 mg CaCO3 was added to the andosol to raise the pH from 5.0 to 6.6. At the start of cycle 3 (end of the oxidizing phase of cycle 2), and for the rest of the treatment, the soil water content of the treated soils was brought to 100 % of the WHC value in an attempt to maximize the decrease in redox potential. No microbial inoculation was conducted. For the initiation of each 3-week reducing phase, the soil of each treated mesocosm was transferred with a shovel into a stainless steel box (120 × 120 × 15 cm) and thoroughly mixed with shovels. The amendments and the tap water required to reach the target water content were distributed at the soil surface. The homogenization of soil and amendment was achieved with shovels and an electric light plough (Gardena®, model EH600/36). The soil was then poured back in the drum. A plastic cover was added at the surface of the soil, and kept in place by ballasts, to minimize possible air entry into the drum that was additionally lidded. For the initiation of the 1-week oxidizing phase following the reducing phase, the transfer of the treated soils from the drum to the stainless steel box and the thorough mixing of the soil were achieved as for the onset of the reducing phase. The soil was then poured back into the drum. Due to the 25 ± 2 °C

9.3 ± 0.6

temperature at which the soils were maintained without the plastic bag and the lid, some desiccation occurred during the oxidizing phase. The required amount of water to be added to reach the water content targeted at the start of the following reducing phase was determined according the ISO standard method 11465 (1994) and applied by spraying over the surface of the soil. The control mesocosms received no amendment during the whole experiment. They were kept at 25 ± 2 °C in the drums for the whole duration of the experiment. The initial water content was maintained by adding the required amount of tap water determined according the ISO standard method 11465 (1994). The experimental design (e.g., drums with 80 kg of soil kept at a controlled temperature of 25 ± 2 °C for 6 months) and the means available for this study did not enable true triplicate of 80-kg mesocosms to be run for each of the three treated and control soils (i.e., a total of 18 mesocosms with 80 kg of soil). The choice was made instead to investigate not one single CLD-contaminated soil type in triplicate but the three major CLD-contaminated soil types (i.e., andosol, nitisol, and ferralsol) with one replicate each. The fact that quartering from an initial soil mass much bigger than the 80 kg needed was used for filling up the treated and control replicate of each soil type contributes to the representativity of the experiment. Measurement of redox potential The redox potential (Eh) was measured using a pH/ORP WTW probe connected to a WTW model 196 multimeter. Readings were taken at a 10-cm depth in three different spots

Environ Sci Pollut Res

of each control and treated mesocosms at two different moments during the 3-week reducing phase of each treatment cycle and at the start and end of the 1-week oxidizing phase. The measured ORP values were converted into Eh values (standard hydrogen potential scale) by adding 220 mV to each ORP determination. Analysis of CLD and its dechlorinated transformation products Triplicate soil samples taken from each mesocosm by driving a stainless steel auger (internal diameter 5.6 cm) over the whole depth of the drum just before the start of the experiment, at the end of the reducing phase of cycles 1, 3, 4, and 6, and at the end of the oxidizing phases of cycles 1, 3, and 6 were used for determination of CLD and possible transformation products. For both phases of cycles 1 and 3, the soil from the 3 cores was pooled to yield one single composite sample used for analysis. At the end of the reducing phase of cycle 4 and at the end of the reducing and oxidizing phases of cycle 6, the 3 cores were analyzed individually. Such pooling for cycles 1 and 3 and the choice not to analyze samples from cycles 2 and 5 were made necessary by limitations in the analytical capacity available for this project. The extraction and analytical methods (Bristeau et al. 2014) are briefly summarized here. The determination of CLD and its transformation products started with Pressurized Liquid Extraction of the 5 g (dw) ground to <80 μm soil sample with a mixture acetone/hexane 50:50 v/v at 100 °C and 110 bars pressure. GC/MS/MS was achieved with a Bruker GC450 gas chromatograph, 1177 injector, automated sampler Combi Pal (CTC) and a triple quadrupole mass spectrometer 300MS. The recoveries from the 3 soil types range from 92 to 139 % (mean = 114 %, standard deviation = 16 %, n = 24) and the quantification limit for CLD is 0.03 mg kg−1 (Bristeau et al. 2014). The dechlorination of CLD can lead to 485 congeners (Dolfing et al. 2012). Only one congener is commercially available as analytical standard, namely 5b-monohydroCLD (CAS number: 53308-47-7, CAS nomenclature: 1,1a,3,3a,4,5,5,5a,5b-nonachlorooctahydro-1,3,4-metheno2H-cyclobuta[c,d]pentalen-2-one; IUPAC nomenclature: 8monohydro-CLD) marketed under the name chlordecone5b-hydro (SM2 Fig. S1). The main CLD transformation product after reaction with ZVI in water/acetone solution and another important one have recently been identified as the 5amonohydroCLD (Belghit et al. 2015) and 5a,6-dihydroCLD (Belghit 2014), respectively (CAS nomenclature). Their chromatographic and mass spectrometry characteristics have been used to unambiguously identify the corresponding products in the present experiments. The 5b,6-dihydroCLD (CAS number: 55570-85-9; 1,1a,3,3a,4,5,5,5a-octachlorooctahydro-1,3,4-metheno-2H-

cyclobuta[c,d]pentalen-2-one) included in libraries of GC/MS (MS search 2.0 NIST 14, Bruker; www.bruker.com) was synthetized by US EPA in the late seventies (Harless et al. 1978) but is no longer available as standard. CLD (ref. C 11220000, purity ≥98 %) and CLD-5b-hydro (ref. LA11220200CY, purity >95 %) were purchased from Cil Cluzeau. The first method used for identifying CLD transformation products consisted in searching the chromatograms for the cyclopentadiene fragments characteristic of CLD (C5Cl6, m/z = 270; m/z = mass-to-charge ratio, where m stands for the mass and z stands for the charge number of the ions, respectively) and its transformation products resulting only from dechlorination (C5Cl6-xHx, 1 ≤ x ≤ 6). The interest and principles of such a method have been established for the determination of pesticides and their degradation products in soil (Andreu and Picó 1991). For mono- and di-hydroCLD, the corresponding information is available in the literature (Harless et al. 1978) The characteristic m/z of all the cyclopentadiene fragments was calculated by the software NIST MS Search 2.0, isotope calculator (http://chemdata. nist.gov/mass-sp/ms-search/). The most intense masses were taken into account (SM3 Table S1). The second method consisted of looking for the masses corresponding to the molecular ion [M+.] and the molecular ion - 1 Cl, [M-Cl+.] of CLD and its 10 dechlorination products (SM3 Table S2). The use of molecular ions for the determination of environmentally important compounds is well documented (Barceló 2004). To optimize sensitivity, the search was conducted in selected ion monitoring (SIM) mode. The m/z of the three most intense molecular ions was calculated using NIST MS Search 2.0 isotope calculator. Identification was accepted when the match between the observed ratio of relative abundances was within 15 % of the theoretical ratio. The relative peak area of each of those substances, calculated according to Eq. 1, was used as a proxy for the concentrations that cannot be determined due to the lack of standards. Eq. 1 was also used for 5a-monohydroCLD and 5a,6dihydroCLD as the standards of these two products became only available 3 years after all other analyses presented here had been conducted. Relative peak area ¼

Areacompound  RC13  V  RM  Df RYtransnano  m  ð1−AMÞ  ð1−RWCÞ

ð1Þ

where: –

Areacompound: area of the chromatographic peak of the compound considered

Environ Sci Pollut Res

– – – – – – – –

RC13: ratio between the known CLDC13 concentration added as internal standard to the soil extract before injection and the CLDC13 peak area measured V: final volume of the soil extract (mL) RM: ratio between the mass of soil extract generated and the mass of extract used for the analysis Df: dilution factor of the final extract before injection RYtransnano: recovery yield of the tracer (transnonachlore) added before the extraction m: mass of soil extracted (g, dried at 38 °C) AM: ratio amendment/soil (w/w) RWC: residual water content of the soil (difference between weight at 105 °C and 38 °C)

(2009). The ISO test based on the germination and growth of two terrestrial plants (ISO standard method 11269-2, 2006) was applied to the soils prepared as recommended by AFNOR standard method NF EN 14735 (2006). Since the possible CLD transformation products were neither known nor available as analytical standards, their individual ecotoxicity could not be studied here.

Statistical analysis Redox values and chlordecone concentrations were analyzed using non-parametric Kruskal-Wallis and/or Mann-Whitney tests. All the statistical analyses were performed in R packages (Thioulouse and Dray 2007; R Development Core Team 2011).

Ecotoxicity assessment

Results and discussion

The limited information available on toxicity of mono- and dihydroCLD compared to that of CLD indicates a decrease in toxicity to laboratory animals as the level of chlorination decreases (Carver and Griffith 1979; Soileau and Moreland 1983). In view of the present research objectives, ecotoxicity tests applied to the soils have been conducted in an independent accredited laboratory (Institut de Recherche Hydrologique, Nancy, France). The ISO tests based on Vibrio fisheri (ISO standard method 11348-1, 2009), Daphnia magna (ISO standard method 6341, 1996), and Brachionus calyciflorus (ISO standard method 20666, 2009) were applied to the control and treated soils at the end of the ISCR treatment. The soil leachates were prepared as recommended by ISO standard method 21268-2 AND

AND/DARA

Cycle

800

Cycle 2

Redox potential In the three control soils, the redox potentials were very similar (Fig. 1) throughout the duration of the study, fluctuating between +500 and +660 mV, except for a period of about 2 weeks covering the end of cycle 1 and start of cycle 2. The addition at cycle 1 of the amendment and 2 % ZVI induced a sharp drop to values ca. −300 ± 30 mV (±, according to the soil) 1 week after the treatment started. The first oxidizing phase yielded in the 3 treated soils Eh values between +554 and +608 mV, just slightly lower than in the controls. FRL

Cycle 3

FRL/DARA

Cycle 4

NIT

Cycle 5

NIT/DARA

Cycle 6

400 200

-200 -400 -600 0

10

20

Oxidizing phase

0

Reducing phase

Redox potential (mV/NHE)

600

30

40

50

60

70

80

90 100 110 120 130 140 150 160 170

Time (days)

Fig. 1 Evolution over time (6 consecutive cycles, each one consisting of 3 weeks anaerobic conditions followed by 1 week oxidizing conditions) of the redox potential measured at 10 cm depth in the control (open

symbols) and treated (full symbols) mesocosms. AND: andosol; FRL: ferralsol; NIT: nitisol; DARA: Daramend® treated

Environ Sci Pollut Res

Overall, the later cycles followed the same trend between phases, with a major difference noted: values during the reducing phases of cycles 2 to 6 never again reached the −300 mV range of the 1st cycle. For each of the three soils, the mean Eh value after the 1st week of the 1st cycle was statistically different from that of the 1st week of each of the following cycles (Mann-Whitney, p value = 0.05). This was rather surprising since an extra addition of ZVZ (albeit limited, 0.03 %) and ZVI (1 % dry weight) was applied at the start of cycle 2 and 3, respectively. The inability of successive additions of strong reducing agents to recreate in soils the initial drop in redox potential after the first addition has been observed, but not explained, in another study for Daramend® combined with ZVI and for ZVI alone (Elgh Dalgren et al. 2009). Here, the hypothesis can be raised that the oxidizing

phase resulted in the added ZVI turning into freshly oxidized iron more readily available for reduction than the initial oxidized species of the soil. As a consequence, the next addition of ZVI or ZVZ would have also to reduce this additional pool of oxidized species, limiting therefore the ability to lower the redox potential to levels as low as those reached when reducing the initial soil alone. The difficulty to generate negative Eh in the tropical soils studied here could be due to their high iron content (7.6– 11.0 %, Table 1), mostly in oxidized state as indicated by the brownish color of the soils. This is supported by literature data (Parfitt et al. 1988; Maejima et al. 2000; Schaefer et al. 2008) establishing that volcanic soils contain iron oxide forms such as Fe(OH)3 (ferrihydrite), α-Fe2O3 (hematite) and αFeOOH (goethite). For the andosol, it can be further

Cycle 1

20

Cycle 2

Cycle 3

Cycle 4

Cycle 5

Cycle 6

18 16 14 12 *

10

6 4 2 0 0

*

* Oxidizing phase

8

Reducing phase

Chlordecone concentration (mg kg-1 of soil)

a

10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 Time (days)

Cycle 1

Cycle 2

Cycle 3

Cycle 4

Cycle 5

Cycle 6

Oxidizing phase

4

3

2 Reducing phase

Chlordecone concentration (mg kg-1 of soil)

b

1

0 0

*

* *

*

* *

10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 Time (days)

Fig. 2 Evolution over time mesocosms with the control symbols). a AND: andosol, b Daramend® treated. Values

of chlordecone concentrations in the (open symbols) and treated soils (full FRL: ferralsol and NIT: nitisol; DARA: at day 0, 105, 163, and 170 are the

mean ± standard deviation, n = 3; for the other sampling dates, n = 2. Asterisk Chlordecone concentration significantly different compared to the control soil (Mann-Whitney test, p value <0.05)

Environ Sci Pollut Res

hypothesized that the very high organic matter content (13.4 %, Table 1) further hampered the reducing effect of the amendments. CLD concentrations The CLD concentrations of the control soils (Fig. 2a, b) fluctuate somewhat over time but with no clear trend. These variations reflect the intrinsic heterogeneity of CLD contamination of the soils. For the treated soils, each time the number of replicates enabled a statistical test to be applied, their CLD concentrations were significantly lower than the control (MannWhitney, p value <0.05). The decrease in CLD concentration at the end of cycle 6 compared to the initial value reached 74 % for the NIT soils and 71 % for the FRL, but was limited to 22 % for the AND (Table 2). Nevertheless, the highest amount of CLD dissipated was observed in the andosol (3.1 mg kg−1) as compared to ferralsol and nitisol (1.8 and 0.9 mg kg−1, respectively). The limited effect of the treatment on the andosol may result from CLD being protected by its inclusion in the nanoporosity of allophones (Woignier et al. 2012) and by its sorption to the organic matter present at particularly high levels in the andosol (Table 1). Albeit limited, the decrease in CLD concentration in the treated andosol, and all the more so for the nitisol and ferralsol, is far greater than the few percents observed in laboratory experiments with CLD in solution and bacteria (3–5 % mineralized after 7 month incubation) and CLD spiked to an andosol and fungi (1.2 % mineralized after 3 weeks) as degrading agents (FernándezBayo et al. 2013; Merlin et al. 2014). Furthermore, field observations and modelling indicate that centuries will be needed for CLD soil concentrations to decrease sharply if only natural processes such as biodegradation and leaching are occurring (Cabidoche et al. 2009). Twenty years after the CLD application was banned, 5b-hydroCLD is the only CLD transformation product frequently quantified in soils from banana plantations (Devault et al. 2016). This CLD dechlorinated transformation product is not formed in our treated soils (see below). Over 20 years, even very slow biodegradation should have led to some formation of 5a-hydroCLD. Lastly, the possible decrease in CLD extractability from the soil as a result of the soil treatment is taken into consideration since the Table 2 Remediation efficiency at the end of the treatment (cycle 6) for the three soils considered.

Soil

Andosol Ferralsol Nitisol a

analytical procedure has been validated for soils with addition of Daramend® (Bristeau et al. 2014). It seems therefore reasonable to assume that the CLD depletion observed here after only 3 weeks resulted mostly from chemical reduction, not microbial degradation. By contrast, the treatment impacted the density and community structure of microorganisms from the soils studied here, especially nitisol and ferralsol (SM4 Table S1 and Fig. S1). CLD dechlorinated transformation products The andosol was the only soil type tested in which 5bmonohydroCLD could be quantified. However, relative peak areas were never higher in the treated soils than in the controls (Table 4), showing that the transformation of CLD by the amendment did not lead to any measurable formation of 5bmonohydroCLD. In addition to 5b-, 5a-monohydroCLD and 5a,6dihydroCLD (CAS nomenclature), other CLD dechlorinated transformation products (for which no analytical standards are available) were identified. For each of these, the most intense cyclopentadiene fragments with their most intense masses and the transition of quantification in MS/MS (selected reaction monitoring mode, SRM) obtained from the mass spectra generated in full scan are presented in Table 3. For the transformation products labeled Bunknown,^ at least one cyclopentadiene fragment characteristic of dechlorinated CLD was detected, but the characteristic mass m/z of molecular ion M+.or molecular ion – 1Cl [M-Cl]+ at the retention time of the corresponding peak was actually not observed. It was therefore decided not to give a definite number of dechlorination to those products. Whenever two compounds show the same level of dechlorination or have both a level than cannot be unambiguously determined, the letters a or b in brackets enable to differentiate one compound from the other. At the end of the treatment, 11 dechlorinated CLD derivatives were found in the andosol, 7 of which being also identified in the ferralsol and nitisol (Table 4). The greater number of peaks observed in the andosol soil may be attributed to the fact that the andosol had a higher initial CLD concentration. This allows for production of a greater mass of transformation products making them easier to identify above the background noise.

Chlordecone concentration (mg kg−1 of soil) Untreated control

After 6 treatment cycles

14.3 ± 1.1 2.6 ± 0.1 1.2 ± 0.2

11.2 ± 0.4a 0.8 ± 0.2a 0.3 ± 0.1a

Decrease in chlordecone soil concentration (%)

22 71 74

Chlordecone concentration significantly different compared to the untreated initial control soil (Mann-Whitney test, p value <0.05)

Environ Sci Pollut Res Table 3 Chlordecone transformation products identified at the end of the treatment (cycle 6) with their retention time, most intense cyclopentadiene fragment(s) with the corresponding most intense masses (m/z) and transition of quantification Chlordecone degradation product

Retention time (min)

Most intense cyclopentadiene fragment(s) and the corresponding 3 most intense masses with relative abundance (% in brackets)

Quantification transition (mode SRM; collision energy 15 V, dwell time 0.02 s)

5a-monohydroCLD

20.6

238 > 203

5b-monohydroCLD

20.4

5a,6-dihydroCLD

19.8

DihydroCLD

19.3

TetrahydroCLD (a)

18.9

TrihydroCLD

18.5

Unknown (a)

18.3

TetrahydroCLD (b)

18.1

PentahydroCLD (a)

18.0

PentahydroCLD (b)

17.1

Unknown (b)

16.4

[C5Cl5H]+.: 236 (62), 238 (100), 240 (64) [C5Cl4H]+: 201 (78), 203 (100), 205 (48) [C5Cl6]+.: 270 (52), 272 (100), 274 (80) [C5Cl5]+: 235 (62), 237 (100), 239 (64) [C5Cl5H]+.: 236 (62), 238 (100), 240 (64) [C5Cl4H]+: 201 (78), 203 (100), 205 (48) [C5Cl4H2]+.: 202 (78), 204 (100), 206 (48) [C5Cl3H2]+: 167 (100), 169 (96), 171 (31) [C5Cl4H2]+.: 202 (78), 204 (100), 206 (48) [C5Cl3H2]+: 167 (100), 169 (96), 171 (31) [C5Cl4H2]+.: 202 (78), 204 (100), 206 (48) [C5Cl3H2]+: 167 (100), 169 (96), 171 (31) [C5Cl4H2]+.: 202 (78), 204 (100), 206 (48) [C5Cl3H3]+: 168 (100), 170 (96), 172 (31) [C5Cl4H2]+.: 202 (78), 204 (100), 206 (48) [C5Cl3H3]+: 168 (100), 170 (96), 172 (31) [C5Cl3H3]+: 168 (100), 170 (96), 172 (31) [C5Cl2H3]+: 133 (100), 135 (64), 137 (10) [C5Cl3H3]+: 168 (100), 170 (96), 172 (31) [C5Cl2H3]+: 133 (100), 135 (64), 137 (10) [C5Cl2H4]+.: 134 (100), 136 (64), 138 (10)

5b-hydroCLD was present in the control of the three soils with a relative peak area representing 7 to 10 % of the CLD peak area. The treatment led to a decrease of 5b-hydroCLD in the 3 soils, especially for the ferralsol and nitisol. 5a-hydroCLD clearly presented the most important relative peak area in the three treated soils (up to 43 % of the CLD peak area in the nitisol), followed by 5a,6-dihydroCLD (up to 22 % of the CLD peak area, in the nitisol again) and by a third transformation product with a loss of 3 chlorides (up to 9 % of the CLD peak area in the ferralsol). A compound with −4 Cl was

272 > 237 238 > 203 204 > 169 204 > 169 204 > 169 204 > 169 204 > 169 170 > 135 170 > 135 134 > 99

observed in the andosol and two compounds with −5 Cl were unambiguously identified in the three treated soils (Table 4), confirming that the decrease in CLD concentrations induced by the treatment comes with extensive dechlorination. The compound Bunknown (b),^ with a retention time 0.7 to 1.6 min shorter than the two −5 Cl (Table 3), was also observed in the three treated soils. Table 3 shows a very clear trend of decreasing retention time as the number of dechlorination increases. It is therefore quite likely that Bunknown (b)^ had a level of dechlorination higher than 5. The fact that

Table 4 Relative intensity (mean ± standard deviation, n = 3) of the peaks of CLD transformation products observed at the end of the treatment (cycle 6) in the control and Daramend® treated (DARA) andosol, ferralsol, and nitisol Compound

CLD 5a-monohydroCLD 5b-monohydroCLD 5a,6-dihydroCLD DihydroCLD TetrahydroCLD (a) TrihydroCLD Unknown (a) TetrahydroCLD (b) PentahydroCLD (a) PentahydroCLD (b) Unknown (b)

Andosol

Ferralsol

Nitisol

Control

DARA

Control

DARA

Control

DARA

73,558 ± 2756 79 ± 6 535 ± 191 0±0

36,573 ± 2544 7601 ± 388 242 ± 100 470 ± 32

8425 ± 982 0±0 65 ± 28 0±0

955 ± 116 332 ± 52 17 ± 2 103 ± 16

4986 ± 660 3±5 53 ± 20 0±0

563 ± 45 243 ± 12 15 ± 4 125 ± 8

152 ± 5 39 ± 4 303 ± 12 17 ± 3 21 ± 3 30 ± 4 52 ± 1 8±2

0±0 0±0 0±0 0±0 0±0 0±0 0±0 0±0

17 ± 3 0±0 84 ± 13 0±0 0±0 10 ± 1 0±0 52 ± 3

0±0 0±0 0±0 0±0 0±0 0±0 0±0 0±0

16 ± 2 0±0 49 ± 43 0±0 0±0 12 ± 3 0±0 40 ± 4

0±0 0±0 0±0 0±0 0±0 0±0 0±0 0±0

Environ Sci Pollut Res

The evolution over time of the relative peak areas of 5ahydroCLD (the greatest relative peak area in the three treated soils) showed a production during the first three cycles, followed by a decline. At the end of the treatment, the relative peak area was for the three soils smaller than the maximum value, up to 10 times smaller for the ferralsol and nitisol (Fig. 3a). The same trend was evidenced for the pentahydroCLD (Fig. 3b) and all the other observed transformation products (data not shown). The relative peak areas of the dechlorinated transformation products were at least 2.3 times (5a-monohydroCLD in the nitisol) and up to 2151 times (unknown (a) in the andosol) lower than that of CLD (Table 4). Therefore, even though the relation between peak areas and concentrations certainly varies according to the compound, it is likely that the concentration of the transformation products was lower than that of the parent compound.

dechlorination of CLD remained apparently incomplete over the 6-month treatment can be explained by the decrease in rate of dechlorination as the number of Cl atoms at the reactive C decreases (Dolfing et al. 2012), and by the fact that strongly negative Eh values (in the −300 mV range) were maintained only during the first 3 weeks. Minute amounts of degradates with – 6 to – 10 Cl could however have been formed but remained undetected. If the dechlorination pathway followed the energetically most favorable steps in a system at equilibrium, only one congener should appear per set of congeners with the same degree of dechlorination, with 5b-monohydro and 3,3adihydro as most likely formed (Dolfing et al. 2012). It is not the case here, since two different mono- and two different penta-hydroCLD are unambigously identified. Additionally, the most abundant monohydroCLD and dihydroCLD formed by the treatment are the 5a-mono and the 5a,6-dihydro, respectively, not the 5b-monohydro and 3,3a-dihydro predicted by thermodynamics. These discrepancies suggest that kinetics played a significant role in the formation of CLD dechlorination products during the treatment.

a

Six of the seven tests applied, including the four terrestrial plant tests of highest relevance in the present case of

AND

Cycle 2

FRL

FRL/DARA

Cycle 4

Cycle 3

NIT

Cycle 5

NIT/DARA

Cycle 6

8000

6000

Oxidizing phase

4000 Reducing phase

5a-monohydroCLD relative peak area

AND/DARA

Cycle 1

10000

2000

0 0

10

20

30

40

50

60

70

80

90

100 110 120 130 140 150 160 170

Time (days)

b

Cycle 1

120

Cycle 2

Cycle 4

Cycle 3

Cycle 5

Cycle 6

100

Oxidising phase

80

60 Reducing phase

pentahydroCLD relative peak area

Fig. 3 Evolution over time of the relative peak area of a 5amonohydroCLD and b pentahydroCLD in the mesocosms with the control (open symbols) and treated soils (full symbols). AND: andosol; FRL: ferralsol; NIT: nitisol; DARA: Daramend® treated. Values at day 105 and 170 are the mean ± standard deviation, n = 3

Ecotoxicity assessment

40

20

0 0

10

20

30

40

50

60

70

80

90

100 110 120 130 140 150 160 170

Time (days)

Environ Sci Pollut Res Table 5 Results (mean of n replicates per leachate concentration, with n = 2 for Vibrio fisheri, n = 4 for Daphnia magna, n = 8 for Brachionus calyciflorus; in brackets: 95 % confidence interval) of the aquatic ecotoxicity tests applied to leachates (liquid/solid ratio = 10/1, 24 h contact) of the three control and the three Daramend®-treated (/DARA) soils at the end of the treatment (cycle 6)

Soil

EC 50 Vibrio fisheri (30 min)

EC 50 Daphnia magna (48 h)

EC 20 Brachionus calyciflorus (48 h)

Non-toxic at 80 % Non-toxic at 80 %

Non-toxic at 90 % Non-toxic at 90 %

43.2 % (16.0–73.3) 40.3 % (23.8–57.2)

Non-toxic at 80 % Non-toxic at 80 %

Non-toxic at 90 % Non-toxic at 90 %

39.4 % (23.3–54.6) 51.9 % (14.3–87.6)

Andosol Control DARA Ferralsol Control DARA Nitisol Control

Non-toxic at 80 %

Non-toxic at 90 %

Non-toxic at 90 %

DARA

Non-toxic at 80 %

Non-toxic at 90 %

52.8 % (32.2–83.4)

agricultural soils, did not show any ecotoxicological response greater in the three treated soils than in the controls (Tables 5 and 6). Only the chronic test based on the rotifer Brachionus calyciflorus applied to the remediated nitisol showed a response greater than the control (Table 5). The greater response observed here for B. calyciflorus compared to the other two target organisms is in agreement with the literature (Radix et al. 2000; Grosell et al. 2006). The toxicity to B. calyciflorus exhibited by the three control soils indicates the presence of toxicants with a mechanism of action efficient towards the test organism. If the toxicant was CLD, one would have expected some differences in toxicity between soils in relation to their very different CLD content (Table 2), which was not the case. It is therefore not clear what caused the toxic effect to B. calyciflorus in the control soils.

Conclusion The CLD concentration decreased by 74 % in the nitisol and 71 % in the ferralsol, resulting more likely from chemical reduction than microbial degradation, and lead to residual

Table 6 Results (mean of 3 replicates; 95 % confidence interval in brackets) of the terrestrial ecotoxicity tests applied to the three control and the three Daramend®-treated (/DARA) soils at the end of the treatment (cycle 6)

Soil

Andosol Control DARA Ferralsol Control DARA Nitisol Control DARA

CLD concentrations of 0.32 and 0.76 mg kg−1, respectively. CLD concentrations below 1 mg kg−1 (dry weight) lead in some edible plants to CLD concentrations lower than the Maximum Residue Level (MRL, 20 μg kg−1 fresh weight; Achard et al. 2007; Cabidoche and Lesueur-Jannoyer 2012). The residual soil CLD concentrations after the remediation of a nitisol at field scale with Daramend® (and zero valent iron alone) lead indeed to CLD concentrations in edible plants lower than in the plants grown on the control soil and lower than the MRL (Mouvet and Bristeau 2016). The remediation of andosol, with generally higher CLD concentrations than the two other soil types (Cabidoche et al. 2009; Cabidoche and Lesueur-Jannoyer 2012), remains a challenge. In the three soil types, the decrease in CLD concentrations was not accompanied by the accumulation of known and detectable CLD-dechlorinated transformation products. The limited data available shows that the toxicity of CLD dechlorinated products is lower than the parent compound (Carver and Griffith 1979; Soileau and Moreland 1983). Furthermore, six ecotoxicity tests, including four on plant germination and growth, did not show any difference between the control and treated soils. The process applied here seems

EC 50

EC 50

Avena sativa

Brassica napus

Germination 12 days

Growth 12 days

Germination 12 days

Growth 12 days

39.0 % Non-toxic at 41 %

> 40 % Non-toxic at 41 %

> 40 % > 41 %

35.9 % (27.1 - > 40) Non-toxic at 41 %

Non-toxic at 78 % Non-toxic at 70 %

Non-toxic at 78 % Non-toxic at 70 %

Non-toxic at 78 % Non-toxic at 70 %

Non-toxic at 78 % Non-toxic at 70 %

Non-toxic at 72 % Non-toxic at 68 %

Non-toxic at 72 % Non-toxic at 68 %

Non-toxic at 72 % Non-toxic at 68 %

Non-toxic at 72 % Non-toxic at 68 %

Environ Sci Pollut Res

therefore promising in not increasing the ecotoxicity of the soils, while decreasing significantly the CLD concentrations, and in producing transformation products with a toxicity lower than that of the parent compound. Work is presently in progress to address the latter perspective. All the procedures applied here at laboratory scale could easily be transposed to field applications. Machinery commonly used in banana plantations can be used to till the soil and ensure mixing of the amendments in the soil. Sprinklers routinely used for irrigating the banana plantations and plastic 50 m2 ground covers readily available in the FWI would enable to reach and maintain the desired soil water content. Field studies remain nevertheless to be conducted to confirm the results and further evaluate if a treatment with Daramend® or other reducing amendments can be applied in a safe and effective manner in situ.

Acknowledgments The help of Dr. Y. M. Cabidoche in selecting the plots for sampling the soils and contacting their owners was greatly appreciated. The input of W. Sowocool was decisive in the identification of the transformation products of chlordecone. Scientific collaboration with Dr. Jim Mueller (PROVECTUS Environmental products, formerly FMC Corporation and Adventus) is gratefully acknowledged. Technical and scientific inputs from Laurent Thannberger (Valgo) were very much appreciated. Thanks for technical assistance to Pierre Gallé-Cavalloni, Pascal Auger, Mickael Beaulieu, Laure Lereau and Hafida Tris. The results presented here were obtained through financing by the French Ministry of Environment (contract 2010 SU 0006693 and 2100598309).

References Abbey AMI, Beaudette LA, Lee H (2003) Polychlorinated biphenyl (PCB) degradation and persistence of a gfp-marked Ralstonia eutropha H850 in PCB-contaminated soil. Appl Microbiol Biotechn 63(2):222–230 Achard R, Perrier X, Chabrier C, Lassoudière A (2003) Cartographie du risque de pollution des sols de Martinique par les organochlorés. Méthodologie d’échantillonnage à la parcelle, Rapport de Phase1; Brgm / RP-52464-FR Achard R, Cabidoche YM, Caron A, Nelson R, Duféal D, Lafont A, Lesueur-Jannoyer M (2007) Contamination des racines et tubercules cultivés sur sol pollué par la chlordécone aux Antilles. Les cahiers du Pram 7:45–50 AFNOR standard method NF EN 14735 (2006) Characterization of waste—preparation of waste samples for ecotoxicity tests Andreu V, Picó Y (1991) Determination of pesticides and their degradation products in soil: critical review and comparison of methods. Trac-Trend Anal Chem 23(10–11):772–789 Baran N, Barras AV (2008) Processus de transfert des produits phytosanitaires du sol vers les eaux souterraines en Martinique. Phase 2: études de processus de sorption et de dégradation dans les sols et phase 3: préconisations de suivi dans les eaux souterraines; BRGM RP56658-FR Barceló D (2004) Applications of gas chromatography-mass spectrometry in monitoring environmentally important compounds. TracTrend Anal Chem 10(10):323–329

Belghit H (2014) Développements analytiques et mécanismes physicochimiques impliqués dans la réduction in situ de la chlordécone dans les sols antillais. PhD Dissertation, University of Orléans (France) Belghit H, Colas C, Bristeau S, Mouvet C, Maunit B (2015) Liquid chromatography – high-resolution mass spectrometry for identifying aqueous chlordecone-hydrate dechlorinated transformation products formed by reaction with zero-valent iron. Int J Env Anal Chem 95(2):93–105 Bertrand JA, Abarnou A, Bocquené G, Chiffoleau JF, Reynal L (2009) Diagnostic de la contamination chimique de la faune halieutique des littoraux des Antilles françaises. Campagnes 2008 en Martinique et en Guadeloupe. Ifremer, Martinique. http://www.ifremer. fr/docelec/doc/2009/rapport-6896.pdf. Accessed 14 Dec 2015 Bocquené G, Franco A (2005) Pesticide contamination of the coastline of Martinique. Mar Poll Bull 51(5–7):612–619 Bristeau S, Amalric L, Mouvet C (2014) Validation of chlordecone analysis for native and remediated French West Indies soils with high organic matter content. Anal Bioanal Chem 406(4):1073–1080 Cabidoche YM, Lesueur-Jannoyer M (2012) Contamination of harvested organs in root crops grown on chlordecone-polluted soils. Pedosphere 22(4):562–571 Cabidoche YM, Achard R, Cattan P, Clermont-Dauphin C, Massat F, Sansoulet J (2009) Long-term pollution by chlordecone of tropical volcanic soils in the French West Indies: a simple leaching model accounts for current residue. Environ Pollut 157(5):1697–1705 Carver RA, Griffith FD (1979) Determination of CLD dechlorination products in finfish, oysters and crustacean. J Agric Food Chem 27(5):1035–1037 Chevallier T, Woignier T, Toucet J, Blanchart E (2010) Organic carbon stabilization in the fractal pore structure of Andosols. Geoderma 159(1–2):182–188 Clostre F, Lesueur-Jannoyer M, Cabidoche YM (2010) Remédiation à la pollution par la chlordécone aux Antilles. http://www.observatoireeau-martinique.fr/les-outils/base-documentaire/conclusions-del2019atelier-abremediation-a-la-pollution-par-la-chlordecone-auxantillesbb. Accessed 14 Dec 2015 Clostre F, Lesueur-Jannoyer M, Achard R, Letourmy P, Cabidoche YM, Cattan P (2014) Decision support tool for soil sampling of heterogeneous pesticide (chlordecone) pollution. Env Sci Poll Res 21(3): 1980–1992 Coat S, Bocquené G, Godard E (2006) Contamination of some aquatic species with the organochlorine pesticide chlordecone in Martinique. Aquat Living Resour 19(2):181–187 Coat S, Monti D, Legendre P, Bouchon C, Massat F, Lepoint G (2011) Organochlorine pollution in tropical rivers (Guadeloupe): role of ecological factors in food web bioaccumulation. Environ Pollut 159(6):1692–1701 Colombano S, Blanc C, Guérin V, Chevrier B (2009) Examen des possibilités de traitement de la chlordécone dans les sols notamment sur les aires d’alimentation des captages d’eau potable, Brgm/RP57708-FR Dallaire R, Muckle G, Rouget F, Kadhel P, Bataille H, Guldner L, et al. (2012) Cognitive, visual, and motor development of 7-month-old Guadeloupean infants exposed to chlordecone. Environ Res 118: 79–85 Devault DA, Laplanche C, Pascaline H, Bristeau S, Mouvet C, Macarie H (2016) Natural transformation of chlordecone into 5bhydrochlordecone in French West Indies soils: statistical evidence for investigating long-term persistence of organic pollutants. Environ Sci Poll Res Int 23(1):81–97 Dolfing J, Novak I, Archelas A, Macarie H (2012) Gibbs free energy of formation of chlordecone and potential degradation products: implications for remediation strategies and environmental fate. Environ Sci Technol 46(15):8131–8139 Elgh Dalgren K, Waara S, Düker A, von Kronhelm T, van Hees PAW (2009) Anaerobic bioremediation of a soil with mixed contaminants:

Environ Sci Pollut Res explosives degradation and influence on heavy metal distribution, monitored as changes in concentration and toxicity. Water Air Soil Poll 202(1–4):301–313 Fernández-Bayo JD, Saison C, Voltz M, Disko U, Hofmann D, Berns AE (2013) Chlordecone fate and mineralisation in a tropical soil (andosol) microcosm under aerobic conditions. Sci Tot Environ 463-464:395–403 Fritz M (2009) L’autorisation du Chlordécone en France 1968–1981, Rapport AFSSET. http://www.observatoire-pesticides. fr/upload/bibliotheque/457291400429630296486151015810 /autorisation_chlordecone_france__1968_1981.pdf. Accessed 14 Dec 2015 Gourcy L, Baran N, Vittecoq B (2009) Improving the knowledge of pesticide transfer processes using age-dating tools (CFC, SF6, 3H) in a volcanic island (Martinique, French West Indies. J Contam Hydrol 108(3–4):107–117 Grosell M, Gerdes RM, Brix KV (2006) Chronic toxicity of lead to three freshwater invertebrates - Brachionus calyciflorus, Chironomus tentans, and Lymnaea stagnalis. Environ Toxicol Chem 25(1):97– 104 Harless RL, Harris DE, Sovocool GW, Zehr RD, Wilson NK, Oswald EO (1978) Mass-spectrometric analyses and characterization of chlordecone in environmental and human samples. Biomed Mass Spectrom 5(3):232–237 ISO standard method 11269-2 (2006) Soil quality—determination of the effects of pollutants on soil flora. Part II: Effects of chemicals on the emergence and growth of higher plants ISO standard method 11348-1 (2009) Water quality—determination of the inhibitory effect on the light emission of Vibrio fischeri (Luminescent bacteria test). Part 1: Method using freshly prepared bacteria ISO standard method 11465 (1994) Soil quality—determination of dry matter and water content on a mass basis—gravimetric method ISO standard method 20666 (2009) Determination of the chronic toxicity to Brachionus calyciflorus in 48 h ISO standard method 21268-2 (2009) Soil quality—leaching procedures for subsequent chemical and ecotoxicological testing of soil and soil materials. Part 2: Batch test using a liquid to solid ratio of 10 L/kg dry matter ISO standard method 6341 (1996) Water quality—determination of the inhibition of the mobility of Daphnia magna Straus (Cladocera, Crustacea)—acute toxicity test Joly PB (2010) La saga du chlordécone aux Antilles françaises. Reconstruction chronologique 1968–2008. Rapport INRA Sens. h t t p : / / w w w. o b s e r v a t o i r e - p e s t i c i d e s . f r / u p l o a d / bibliotheque/852173530783222242256849728077/saga_ chlordecone_antilles_francaises_1968_2008.pdf. Accessed 14 Dec 2015 Jondreville C, Lavigne A, Jurjanz S, Dalibard C, Liabeuf JM, Clostre F, Lesueur-Jannoyer M (2014) Contamination of free-range ducks by chlordecone in Martinique (French West Indies): a field study. Sci Tot Environ 493:336–341 Journal Officiel de la République Française (2005) Arrêté du 5 octobre 2005 relatif à la teneur maximale en chlordécone que ne doivent pas dépasser certaines denrées d’origine animale pour être reconnues propres à la consommation humaine. http://www.observatoirep e s t i c i d e s . g o u v. f r / u p l o a d / b i b l i o t h e q u e / 8 9 0 9 50301230666866433863582907/arrete-teneur-maximalechlordecone-5oct05.pdf. Accessed 14 Dec 2015 Kadhel P, Monfort C, Costet N, Rouget F, Thomé JP, Multigner L, Cordier S (2014) Chlordecone exposure, length of gestation and risk of preterm birth. Am J Epidemiol 179(5):536–544 Kim SC, Yang JE, Ok YS, Sik YO, Skousen J, Kim DG, Joo JH (2010) Accelerated metolachlor degradation in soil by zero-valent iron and compost amendments. Bull Environ Contam Toxicol 84(4):459– 464

Maejima Y, Nagatsuka S, Higashi T (2000) Mineralogical composition of iron oxides in red-and yellow-colored soils from southern Japan and Yunnan, China. Soil Sci Plant Nutr 46(3):571–580 Mercier A, Dictor MC, Harris-Hellal J, Breeze D, Mouvet C (2013) Distinct bacterial community structure of 3 tropical volcanic soils from banana plantations contaminated with chlordecone in Guadeloupe (French West Indies). Chemosphere 92(7):787–794 Merlin C, Devers M, Crouzet O, Heraud C, Steinberg C, Mougin C, Martin-Laurent F (2014) Characterization of chlordecone-tolerant fungal populations isolated from long-term polluted tropical volcanic soil in the French West Indies. Environ Sci Poll Res 21(7):4914– 4927 Monti D, Coat S (2007) La contamination des espèces d’eau douce. Les Cahiers du Pram 7:29–33 Moser H, Roembke J, Donnevert G, Becker R (2011) Evaluation of biological methods for a future methodological implementation of the hazard criterion H14 Becotoxic^ in the European waste list (2000/532/EC). Waste Manag Res 29(2):180–187 Mouvet C, Bristeau S (2016) Comparaison du transfert sol-plantes entre la chlordécone et ses produits de dégradation formés par déchloration réductive. Rapport final. BRGM/RP-65275-FR, 50 p., 11 fig., 15 tabl. http://infoterre.brgm.fr/rapports/RP-65275-FR. pdf. Accessed 2 June 2016 Multigner L, Ndong JR, Giusti A, Romana M, Delacroix-Maillard H, Cordier S, et al. (2010) Chlordecone exposure and risk of prostate cancer. J Clin Oncol 28(21):3457–3462 OPECST (Office Parlementaire d’Evaluation des Choix Scientifiques et Technologiques 2009) Rapport sur les impacts de l’utilisation de la Chlordécone et des pesticides aux Antilles : bilan et perspectives d’évolution, par M. Jean-Yves LE DÉAUT, député et Mme Catherine PROCACCIA, sénateur. http://www.senat.fr/rap/r08-487 /r08-4871.pdf. Accessed 14 Dec 2015 Pandard P, Devillers J, Charissou AM, Poulsen V, Jourdain MJ, Férard JF, et al. (2006) Selecting a battery of bioassays for ecotoxicological characterization of wastes. Sci Total Environ 363(1–3):114–125 Parfitt RL, Childs CW, Eden DN (1988) Ferrihydrite and allophane in four Andepts from Hawaii and implications for their classification. Geoderma 41(3–4):223–241 Phillips TM, Lee H, Trevors JT, Seech AG (2004) Mineralization of hexachlorocyclohexane in soil during solid-phase bioremediation. J Ind Microbiol Biotechnol 31(5):216–222 Phillips TM, Seech AG, Lee H, Trevors JT (2005) Biodegradation of hexachlorocyclohexane (HCH) by microorganisms. Biodegradation 16(4):363–392 Phillips TM, Lee H, Trevors JT, Seech AG (2006) Full-scale in situ bioremediation of hexachlorocyclohexane-contaminated soil. J Chem Technol Biotechnol 81(3):289–298 Préfecture de la Région Guadeloupe (2010) Recueil des actes administratifs. Arrêté n° 2010–721 PREF/DSV du 23 juin 2010 réglementant la pêche et la commercialisation des espèces de la faune marine dans certaines zones maritimes de la Guadeloupe. http://www.guadeloupe.pref.gouv.fr/content/download/3088/17231 /file/Arrete_prefectoral_du_23_juin_2010_reglementationchlordecone-peche.pdf. Accessed 14 Dec 2015 R Development Core Team (2011) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3–900051–07-0, URL. http://www.Rproject.org. Accessed 10 Sept 2015 Radix P, Leonard M, Papantoniou C, Roman G, Saouter E, GallottiSchmitt S, et al. (2000) Comparison of four chronic toxicity tests using algae, bacteria, and invertebrates assessed with sixteen chemicals. Ecotoxicol Environ Saf 47(2):186–194 Römbke J, Moser T, Moser H (2009) Ecotoxicological characterization of 12 incineration ashes (MWI) using 6 laboratory tests. Waste Manag 29(9):2475–2482

Environ Sci Pollut Res Schaefer CEGR, Fabris JD, Ker JC (2008) Minerals in the clay fraction of Brazilian Latosols (Oxisols): a review. Clay Miner 43:137–154 Seech A, Bolanos-Shaw K, Hill D, Molin J (2008) In Situ bioremediation of pesticides in soil and groundwater. Remediation 19(1):87–99 Soileau SD, Moreland DE (1983) Effects of chlordecone and its alteration products on isolated rat liver mitochondria. Toxicol Appl Pharmacol 67(1):89–99 Thioulouse J, Dray S (2007) Interactive multivariate data analysis in R with the ade4 and ade4TkGUI packages. J Stat Soft 22(5):1–14 UNEP/POPS/POPRC.1/10 (2005) Stockholm Convention on Persistent Organic Pollutants. Persistent Organic Pollutants Review Committee. First meeting. Geneva, 7–11 November 2005. http://www.pops.int/documents/meetings/poprc/meeting_ docs/reports/report_E.pdf. Accessed 14 Dec 2015 UNEP/POPS/POPRC.3/10 (2007) Projet d’évaluation de la gestion des risques : chlordécone. http://www.pops.int/documents/

meetings/poprc_3/meetingdocs/poprc3_doc/10/K0762894.F_ POPS_POPRC_3_10.pdf. Accessed 14 Dec 2015 United States Patent (2000) Composition and method for dehalogenation and degradation of halogenated organic contaminants, US Patent N° 5, 618, 427 Woignier T, Fernandes P, Jannoyer-Lesueur M, Soler A (2012) Sequestration of chlordecone in the porous structure of an andosol and effects of added organic matter: an alternative to decontamination. Eur J Soil Sci 63(5):717–723 Woignier T, Fernandes P, Soler A, Clostre F, Carles C, Rangon L, Lesueur-Jannoyer M (2013) Soil microstructure and organic matter: keys for chlordecone sequestration. J Hazard Mater 262:357–374 Zhuang L, Gui L, Gillham RW, Landis RC (2014) Laboratory and pilotscale bioremediation of pentaerythritol tetranitrate (PETN) contaminated soil. J Hazard Mater 264:261–268