The fate of organic micro-pollutants in soil largely depends on their interactions with soil organic matter (SOM). Indeed, the intensity and nature of such interactions are major factors in the control of micro-pollutant bioavailability, and hence degradation and micro-pollutant mobility in the environment [Khan and Hamilton, 1980 ; Calderbank, 1989 ; Dec et al., 1990, 1997 ; Barriuso and Koshinen, 1996]. Residual micro-pollutants, i.e. the remaining micro-pollutants and their various transformation products (metabolites), occur in soil both in the water phase and associated with the solid phase. A part of these associated molecules, termed bound residues (BRs), cannot be released via extraction with water or organic solvents like methanol. SOM plays a major role in BR formation [e.g. Loiseau et al., 2000] and different types of interaction (like covalent bonds and trapping in organo-mineral aggregates) can be implied [Hsu and Bartha, 1976 ; Schiavon et al., 1977 ; Bollag et al., 1992 ; Dec and Bollag, 1997 ; Steinberg et al., 1987; Providenti et al., 1993].
The present study is concerned with the fate of atrazine in three silty loamy soils from the Paris Basin. Atrazine is largely used (ca. 5000 T/year in France) as herbicide for maize cropping and may cause important pollution problems through transfer to water resources.
Incubations were performed for 60 days, using 14C labelled atrazine, with the selected soils. These incubations were carried out with 50 g of soil in closed vessels at room temperature in darkness. The production of 14C CO2 was regularly measured so as to determine the rate of atrazine mineralisation in the three incubated soils. These three soils exhibit large differences in pH and atrazine use (tabl. I). Following incubations, radioactivity measurements were performed on the different fractions isolated : aqueous extract, methanol extract, BRs in various size fractions (< 20 μm, 20–50 μm, > 50 μm). In addition, BR distribution was determined in the fulvic and humic acids and humin isolated from the < 20 μm fraction. In a second set of incubation experiments, the distribution of residual atrazine was examined in size fractions of organo-mineral aggregates separated by sieving in water (< 50 μm, 50–200 μm, 200 μm-1 mm, 1–2 mm, > 2 mm) and in free plant debris separated by floating. The GB1 soil was also incubated for 60 days using 13C labelled atrazine. Curie point flash pyrolysis combined both with isotopic measurements on individual compounds (Py-GC-C-irMS) and with gas chromatography/mass spectrometry (GC-MS) analyses was used to examine atrazine BRs following the latter incubation.
Results and discussion
The extent of atrazine mineralisation during incubation is sharply different for the three soils (fig. 1). The GM1 soil shows an intense mineralisation and ca. 85 % of the applied atrazine is transformed into CO2. Moreover, this degradation is fast and mostly takes place during the first week. In contrast, much lower levels of mineralisation are observed for the other two soils, especially for VMSA (only ca. 8 % after 60 days). For the GB1 soil, weak mineralisation is noted, during the first week of incubation. However, following this lag phase, substantial mineralisation occurs and a value of ca. 20 % is obtained at the end of the experiment. It was previously observed [Houot et al., 1998] that soil microflora can adapt to atrazine mineralisation following regular inputs of atrazine. Accordingly, the pronouced differences observed in the present study reflect the presence of a microflora adapted to atrazine degradation in the GM1 soil, due to yearly application of atrazine over 25 years whereas such a microflora is absent in the GB1 and VMSA soils, which had never been treated with atrazine before incubation. The substantial difference in mineralisation intensity observed between the latter two soils is probably related to the relatively low pH of the VMSA soil since such a feature is known to limit atrazine mineralisation [Houot et al., 1998 ; Castéraz, 1998].
BRs and distribution of residual atrazine
The distribution of residual atrazine between the different fractions controls its biodisponibility and fate. Water-soluble molecules are considered as directly bioavailable, whereas the degree of availability is much lower for the bound compounds and, to a lesser extent, for the methanol-soluble ones. It appears (fig. 2) that, as expected, the relative abundance of water– and methanol-soluble residual atrazine is inversely correlated to mineralisation intensity. As a result, the contribution of BRs to total residual atrazine is much higher for GM1 (68 %) where soluble residues were strongly affected by mineralisation when compared to GB1 (36 %) and VMSA (51 %). The higher percentage observed for VMSA relative to GB1 suggests that BR formation is favored by a low pH.
BR distribution between the different size fractions (fig. 2) shows that the largest absolute amount of BRs is found for the three soils in the < 20 μm fraction. This fraction is both the most humified and the most abundant (as wt %). However, if BR concentration relative to OM is considered (tabl. II), it appears that the highest concentrations in BRs tend to occurs in the fractions which contain non- (or weakly-) humified organic material. Atrazine BRs thus show a higher affinity with the latter material, as also previously observed for various soils [Barriuso et al., 1991 ; Barriuso and Koshinen, 1996 ; Loiseau et al., 2000].
BR distribution between the fulvic and humic acids and humin was examined for the < 20 μm fraction (fig. 3). For the three soils, the largest amount of BRs is associated with humin, especially in the case of GB1 and GM1. Substantial amounts of BRs are also found with the fulvic acid fractions. Moreover, when BR concentration is considered (tabl. III), a higher affinity is noted for fulvic acids. The latter features should have important environmental consequences since fulvic acids are rather readily transfered from soil to surface and ground water. The high affinity of atrazine BRs for fulvic acids is probably related to interactions with the polar groups that abundantly occur in the latter fraction [Khan and Hamilton, 1980 ; Bertin and Schiavon, 1989 ; Capriel and Haisch, 1983 ; Capriel et al., 1985].
Comparison of the three soils therefore shows pronounced differences in mineralisation levels, related to microflora adaptation and pH. Large differences are also observed in the distribution of residual atrazine between the BRs and the soluble fractions. In contrast, the three soils exhibit similar features for BR distribution (absolute amounts and concentrations) between the three size fractions and also between the fulvic acids, humic acids and humin of the < 20 μm fraction. Concentration values show higher affinity of atrazine BRs for weakly humified OM and for fulvic acids.
Atrazine distribution in organo-mineral aggregates and fresh debris of plants
This study was performed on the GM1 and VMSA soils. Measurements were performed (i) at time zero (just one hour after the addition of 14C atrazine to soils) i.e. when no significant atrazine mineralisation and transformation had occurred and (ii) at the end of the experiment after 60 days of incubations.
Distribution at time 0 is similar for the two soils (fig. 4a) and the bulk of the atrazine is mostly associated with the 0.2–1 mm fraction followed by the 1–2 mm fraction. When content is considered (fig. 4b), the highest value is observed, by far, for the free plant debris. Such a feature reflects, as observed in previous studies [Barriuso et al., 1994 ; Puget et al., 2000], the high affinity of atrazine for fresh or weakly humified material. These results are therefore consistent with the preferential association of atrazine and of its metabolites with weakly humified OM observed in the BRs in the first set of incubations.
After 60 days, the amount of residual atrazine observed in a given fraction shall both reflect the extent of mineralisation in this fraction and possible transfer from other fractions. As already stressed, only weak mineralisation occurs in the VMSA soil and similar atrazine distributions are noted for both soils at time 0. Thus, the results obtained after incubation shall mostly reflect the transfer processes for the VMSA soil and a combination of such processes and of biodegradation for GM1. Accordingly, the relative changes observed, between the two soils, for label concentration in the different fractions (fig. 4c) reflect the relative efficiency of the stabilisation of residual atrazine. Large (ca. 90 %) and similar decreases are observed for all the fractions of organo-mineral aggregates whereas only a limited decrease of ca. 15 % occurs for the free debris. Therefore residual atrazine stabilisation is more important in the latter fraction than in the organo-mineral aggregates and no preferential stabilisation takes place in some size fractions of these aggregates.
Pyrolytic studies of atrazine BRs
Various types of interactions between bound residual atrazine and SOM, ranging from relatively weak linkages like hydrogen bonds to much stronger linkages like covalent bonds, can occur in soil. The nature of such interactions should have major consequences for the stabilisation and fate of residual atrazine. Information on this nature could be derived from successive pyrolyses at increasing temperatures, as recently shown for pyrene retention in sediments [Guthrie et al., 1999] and 2-aminobenzothiazole interactions with humic acids [Schulze et al., 1997]. This type of pyrolytic study, combined with GC/MS identifications, should thus be useful to decipher (i) the structure of the different metabolites in bound residual atrazine, (ii) their relative abundance and (iii) their modes of linkage with SOM. However, this is made difficult by the extremely high dilution of bound residual atrazine both by soil minerals and SOM and it seems that no such study has been reported so far. The above drawbacks could be overcome via incubation with labelled atrazine followed by elimination of the bulk of soil minerals.
To test the suitability of this approach, the GB1 soil was incubated for 60 days with 13C atrazine. After incubation, soluble compounds were eliminated by water and methanol extractions and minerals via a mild treatment with 2 % HF which should not largely affect organic components in soil [Skjemstad et al., 1994]. Atrazine BRs were then examined by pyrolysis combined both with GC-C-irMS and GC-MS. The pyrolysis products derived from bound residual atrazine can be located by isotopic analysis owing to their very high enrichment in 13C (in the 100-1 000 ‰ range) (fig. 5a). As expected, when the GC-MS trace is considered (fig. 5b), the latter products are overwhelmed by the pyrolysis products from the natural OM of the soil. However, they can be identified and analysed owing to the above isotopic localisation and to the presence of typical fragments of the triazine ring (such as m/z 68) in their mass spectra. The retention times and mass spectra of the five compounds derived from bound residual atrazine thus identified indicate that they do not correspond either to atrazine itself or to the metabolites, like hydroxy– and dealkylatrazines, classically observed in the soluble phase. Residual atrazine occurring as BRs and as soluble components in the incubated soil thus appears to exhibit different chemical composition. These preliminary results confirm the suitability and interest of such a combined pyrolytic approach, applied for the first time to the best of our knowledge, for the study of BRs of micro-pollutants.