Response of sorption processes of MCPA to the amount and origin of organic matter in a long‐term field experiment

Changes in farming practices over long times can affect the sorption behaviour of MCPA ((4‐chloro‐2‐methylphenoxy)acetic acid). We studied the adsorption–desorption mechanisms of MCPA on soil with varied amounts and origins of soil organic matter obtained from a long‐term field experiment with various organic amendments. The origin of the soil organic matter seems to be crucial for the sorption behaviour of MCPA. Samples of soil amended with sewage sludge sorbed MCPA more strongly than the soil under any other treatment. Peat‐amended soil was second followed by soil receiving animal manure, green manure, mineral fertilizer without N and the fallowed soil. Both the carbon content and the origin of the organic matter are important for the sorption. A decrease of carbon content of a soil does not necessarily imply a reduction of sorption capacity for polar organic acids such as MCPA. Nevertheless, our adsorption–desorption experiments suggest that with decreasing carbon content the role of mineral sorption mechanisms could become more pronounced. Our results showed that interactions of soil organic matter and soil minerals distinctly influence adsorption properties for MCPA.


Introduction
The leaching of pesticides into groundwater is receiving increasing attention. Adsorption on the soil matrix appears to be one of the most important factors affecting both leaching and behaviour and fate of pesticides in soil. Adsorption± desorption reactions of pesticides in soil are affected by the physicochemical characteristics of the pesticide, the composition and texture of the soil, and temperature and humidity.
The contribution of the individual factors to the possible migration of pesticides to the groundwater can be determined fundamentally by studying the adsorption, desorption, degradation and percolation (Socõ Âas-Viciana et al., 1999). The pesticide sorption behaviour is related both to texture and properties of the soils and to the chemical structure of the organic substances.
Phenoxyalkanoic acid herbicides such as MCPA ((4-chloro-2-methylphenoxy)acetic acid), are broad-spectrum, post-emergent herbicides much used in agriculture. They have one polar carboxylic group and one lipophilic phenyl moiety, and because of this structure they are suitable model substances for studying adsorption±desorption processes in soil (Celis et al., 1999;Socõ Âas-Viciana et al., 1999;Haberhauer et al., 2000). Parker & Rate (1998) postulated that the interactions of the carboxylic groups with surfaces of organic matter and negatively charged clay via metal ion bridges dominate the sorption of acid herbicides. Benoit et al. (1996) thought that partition via lipophilic interactions in soil organic matter dominates. Soil pH can affect the protonation±deprotonation reaction of ionizable organic compounds, and therefore it can in¯uence their sorption behaviour (Nicholls & Evans, 1991).
We believe that ionic interactions of carboxylic acids dominate sorption of phenoxyalkanoic acid herbicides on clay; nevertheless other reactions may also play a role in the sorption and leaching of acid pesticides in soil. Hydrogen bonds, van der Waals interactions of aliphatic groups and charge transfer interactions of the aromatic ring systems of humic substances with aromatic groups of pesticides may affect the strength of adsorption on soil constituents (Dec & Bollag, 1997;Gevao et al., 2000). Non-reversible and non-equilibrium conditions ± within the timescale of the experiment ± (e.g. slow diffusion into and partitioning in the soil matrix) can produce hysteresis when pesticides desorb.
However, the mechanism involved in such sorption has not yet been clearly demonstrated, one of the major reasons being the dif®culty of isolating unaltered organic matter from the soil and characterizing its chemical and physical properties (Benoit et al., 1996).
Hence, the origin, composition and content of the organic matter has an important impact on the sorption behaviour of the soil (Kile et al., 1999;Perminova et al., 1999). The content of organic matter of agricultural soils is related to farming practices including manuring. Thus, changes in farming practices on a long timescale will change organic matter content and therefore may change the adsorption±desorption properties for pesticides.
One long-term experiment was started on a homogeneous ®eld 44 years ago in Sweden. Originally, all plots had the same mineral and organic composition. After adding different amendments over more than 35 years the soil organic matter, aggregate stability and pore sizes have been signi®cantly altered (Gerzabek et al., 1995;Kirchmann & Gerzabek, 1999). All these factors in¯uence the adsorption of pesticides. Thus, long-term experiments supply samples of soil with differing amounts and qualities of organic matter (SOM), but all samples come from the same original soil. Adsorption experiments using such soils may become representative of the real environmental behaviour of the organic matter in agricultural soils subjected to pesticide treatments. Additionally, sorption depends on the nature and amount of surface ultimately exposed (Celis et al., 1999). Thus, laboratory experiments in which certain amounts of organic matter are added to soil samples may give less conclusive results concerning the impact of such amendments than sorption experiments of soil with varying amounts of organic matter obtained from long-term trials.
The objective of this study was to examine the adsorption± desorption behaviour of MCPA as a model substance for anionic herbicides on soil to which had been added various organic materials over some 40 years, and to compare the sorption behaviour with respect to soil organic matter content and origin.

Chemical
Analytical pure MCPA (4-chloro-2-methylphenoxy)acetic acid; Dr Ehrenstorfer GmbH, Augsburg, Germany) was used as analyte. Its solubility in water at 25°C was 825 mg l ±1 , and its acid constant (pK A ) is 3.07. The agronomic dose of MCPA is between 1 and 2.5 kg ha ±1 .

Soil
The soil was from a long-term experiment started in 1956 at Uppsala, Sweden on a Eutric Cambisol (FAO) with 37% clay and 41% silt. The parent material consists of post-glacial clay with illite as the main clay mineral. Kirchmann et al. (1994) document the experiment and data fully. In 1956 the soil (0±20 cm depth) had 15 g kg ±1 of organic carbon, 1.7 g kg ±1 of nitrogen and a pH of 6.6. The experimental design is based on 14 differing treatments, laid out in a randomized block design with four replicates. The individual plots (2 m Q 2 m) were separated by pressure-treated wooden frames. Six of the treatments, namely fallow (weeds removed), NoN (crop rotation with mineral P and K but without nitrogen fertilizer), green manure, animal manure (well decomposed), peat (sphagnum) and sewage sludge, were selected for this study. The application of organic amendments was based on equal amounts of ash-free organic matter amounting to 2000 kg C ha ±1 year ±1 on average. Organic matter was added in autumn in 1956,1960,1963 and thereafter every second year by hand. All plots received a dressing of 20 kg P ha ±1 in the form of superphosphate and 35±38 kg K ha ±1 in the form of potassium chloride annually in spring. Cereals (70%), rape (25%) and fodder beet (5%) were grown in rotation. There were no legumes in the rotation. At harvest the above-ground portion of the crop was completely removed. Samples of topsoil (0± 20 cm) were taken in autumn 1998 from two replicates of each treatment. They were air-dried, passed through a 2-mm sieve, and then stored. Selected soil properties are shown in Tables 1 and 2. Particle-size distribution was determined by the method of Stemmer et al. (1998). This physical fractionation uses lowenergy ultrasonication to disperse the soil and thus preserves stable microaggregates in the silt fraction. Total carbon (= total organic carbon) was measured in the soil samples using an elemental analyser (Carlo Erba Nitrogen Analyser 1500, Milano).

Adsorption experiments
Experimental batch sorption data were obtained as follows: 2 g of each soil sample was weighed into 50 ml centrifuge tubes made of polypropylene copolymer. Five millilitres of 0.01 M CaCl 2 aqueous solutions of MCPA containing initial pesticide concentrations (C 0 ) ranging between 0.1 and 100 mg l ±1 was added to each. The¯asks were shaken for 24 h. Preliminary kinetic experiments had demonstrated that adsorption equilibrium was reached within less than 24 h. A 0.01 M solution of CaCl 2 was used to ease phase separation and to keep ionic strength similar to natural soil solutions. The experiments were carried out at 20°C. After centrifuging, the supernatant solution was removed and the concentrations of MCPA in the supernatant solution, C s , were determined by HPLC with a UV-detector (Beckman Instruments, Fullerton). As mobile phase a methanol±water mixture (ratio 65:35) was used with 0.5% acetic acid (isocratic elution). The analysis was done on a Hypersil 5 "m ODS, 250 mm Q 4.0 mm column (C18material, Merck, Darmstadt, Germany) at 40°C and a¯ow of 1 ml minute ±1 . The MCPA was analysed at the wavelength of maximum absorption (228 nm). The MCPA adsorbed, C a , was calculated from the difference between the initial and ®nal concentration of the CaCl 2 solution. Blanks containing no MCPA were included, and¯asks containing only MCPA and no soil were analysed to determine adsorption on the surface of the¯asks. All sorption experiments were done at least in duplicate. Linear isotherms were obtained (C a = K D C s ) with correlation coef®cients always greater than 0.999. Distribution coef®cients, K D values, were obtained from the slope of the graph of C a against C s . Values of K OC were calculated as follows:

Desorption experiments
Desorption experiments were done in two steps and commenced immediately after the corresponding adsorption experiment. The supernatant removed was replaced by the same amount (5 ml) of fresh 0.01 M CaCl 2 solution. The mixture was shaken for 24 h, and the soil and aqueous phases were separated as described above. This procedure was repeated in a second desorption step. All desorption experiments were performed at least in duplicate.
The MCPA adsorbed, C a-Des1 , C a-Des2 , after the ®rst and second desorption step was calculated from the difference between the initial C 0 and ®nal concentration of the CaCl 2 solution of the adsorption, C s , ®rst desorption, C s-Des1 , and second desorption step, C s-Des2 , respectively. Linear isotherms were obtained by plotting the corresponding adsorbed MCPA after desorption step against the concentration of MCPA in solution: Distribution coef®cients, K D values, were obtained from the slope of a C a-Des1 against C s-Des1 .

Theoretical calculations
We made the initial calculations using the self-consistent ®eld (SCF) Hartree±Fock method (Roothaan, 1951) and the 3-21G basis set (Binkley et al., 1980) as implemented in Gaussian 98 (Gaussian, Inc., Pittsburgh, PA). The interaction energies between MCPA and the individual minerals and model compounds of the soil organic matter (Figure 3) were obtained as follows: ®rst, the energies of minimized conformations of the single molecules were calculated, then the energies of minimized conformations of the whole complexes (MCPA plus model compound) were calculated. The formation energies of these complexes were obtained by subtracting the energies of the single molecules (residues) from the energies of the complex. This formation energy was used as an estimate of the interaction energy.

Results and discussion
After more than 40 years of the treatments the amount (Tables 1 and 2) and composition of soil organic matter has been altered (Gerzabek et al., 1995(Gerzabek et al., , 1997Kirchmann et al., 1996).
The organic carbon contents of the topsoils changed distinctively during the 42 years of treatment. The topsoil on the continuous fallow plots had lost approximately one third of its organic matter. Arable cropping resulted in a less dramatic decrease of organic carbon (13%), and green manure maintained the humus status since 1956. In comparison with the other variants the carbon content has been increased signi®cantly by peat and sewage sludge amendments. The organic matter content remained fairly constant in the soil of plots treated with green manure and increased in the plots receiving animal manure.
The C/N ratio in the topsoil ranged from 9.7 to 18.3 in 1998 (Table 1). The soil under all treatments except the peat had C/N ratios that are typical for agricultural soils (Scheffer & Schachtschabel, 1998). There were signi®cant changes in the characteristics of the organic material extracted from the soil treated with peat, which differed considerably in its E 4 /E 6 ratios, the ratios of fulvic to humic acids, and FTIR-spectra

Sorption responses of MCPA to organic amendments 281
from that under all other treatments (see Gerzabek et al., 1997).

Adsorption
The adsorption of MCPA on the different treatments in batch experiments revealed the differences induced by farming practices. Figure 1 shows adsorption isotherms. There were signi®cant differences in the adsorption behaviour between the peat, sewage sludge, animal manure plots and the other treatments. The soil amended with sewage sludge adsorbed more strongly than on soil treated with peat (PE). The soil that received animal manure (AM) adsorbed slightly more strongly than that to which green manure was added (GM), the unfertilized soil (NoN) and the fallowed soil (FA). Concerning adsorption of MCPA, no signi®cant difference was observed between GM, NoN and FA. The adsorption distribution coef®cients, K D-Ads , obtained from a linear adsorption model ranged from 0.86 to 2.12 for FA and SE treatments, respectively. We were surprised to ®nd no signi®cant Figure 1 Adsorption and desorption isotherms of MCPA on soil of different treatments. The amount of MCPA sorbed, C a , is plotted against the amount in solution, C s , of the adsorption step (ads) and of the ®rst (des1) and second desorption step (des2). correlation between the total carbon content of the samples and their MCPA adsorption behaviour. The K D-Ads value increases from GM, AM to peat and sewage sludge (Table 3 and Figure  2), which accords with increasing carbon content, but the nearly equal K D-Ads of GM, NoN and FA does not correlate with the decrease of organic carbon content from 1.8% (GM) to 1.1% (FA). The fallowed soil adsorbed more strongly than the soil under treatments GM and AM when the large differences in organic matter are taken into account. Additionally adsorption of MCPA on the peat-amended soil was less than that on receiving sludge, although the carbon content is distinctly greater. These ®ndings can be emphasized by calculating K OC values. Nearly the same K OC values (70±80) were obtained for the soil of the fallow and sludge treatments (Table 2), while the K OC was only 50 for the AM, GM and PE treatments. A comparison of the adsorption behaviours of the soil that received peat, which contains the most organic carbon, and that amended with sludge suggests that the origin of the organic matter is crucial for the ability to interact with MCPA. It is obvious that an increase in carbon content in soil alone does not necessarily result in greater adsorption, since the number of adsorption sites for interactions between MCPA and the organic matter depends on the origin of the material (sewage sludge or peat). Organic matter originating from peat, which is a lipophilic and inert amendment (Gerzabek et al., 1997), is less polar and therefore has fewer sites for polar sorption with MCPA than organic matter from sludge. The sludge provides additional adsorption sites for MCPA, as demonstrated by our sorption experiments. The large K OC value of the fallowed soil (least organic matter) and NoN variants might be explained both by variations in polarity of the organic matter (Kile et al., 1999) and by an additional contribution to the sorption of the inorganic soil constituents. We know that clay minerals have reactive surfaces that are normally coated by organic matter (Parker & Rate, 1998). The smaller amount of organic matter as found especially in the fallow plots might make these surface edges more accessible to direct interactions between MCPA and minerals. This explanation accords with the macroaggregates (1±2 mm) being less stable (Gerzabek et al., 1995) and containing more clay-sized particles in the fallow soil ( Table 2).
The degree of dissociation of organic acids such as MCPA affects their sorption behaviour (Nicholls & Evans, 1991). The pH of the soil amended with peat and sludge (pH = 5.8) is less than that of the soil under the other treatments (6.3±6.7) but still more than two orders of magnitude greater than the pK A value of MCPA (pK A = 3.07). However, the pH of the soil solution near soil surfaces decreases. A larger portion of the neutral form of MCPA may increase partitioning of MCPA in the lipophilic soil matrix. Thus, besides containing more organic matter the somewhat greater acidity of the soil amended with peat and sludge might contribute to the greater adsorption of MCPA in comparison with the other treatments.

Desorption
The desorption behaviour of MCPA was similar to the adsorption behaviour. The largest K D-Des values were obtained for the soil treated with peat and sludge followed by the FA, NoN and AM treatments ( Table 3). The quotient of the K D-Ads / K D-Des1 shows the hysteresis of the sorption process. While   similar quotients were calculated for AM, GM, PE and SE (0.61±0.67) distinctly smaller quotients were obtained for the NoN and even smaller ones for the fallow variant. Desorption relative to adsorption of MCPA is least favoured in soil from the fallowed plots. Such hysteresis is commonly found in sorption experiments. For example, diffusion-limited processes and strong or irreversible binding are thought to contribute to the hysteresis (Pignatello & Xing, 1996).
A set of possible molecular interactions of MCPA with mineral or organic matter is shown in Figure 3. The binding sites considered consist of either a mineral moiety, which can be regarded as a model for a mineral surface edge, or an organic molecule representing a structural moiety of SOM and MCPA (Figure 3). Both interactions of the neutral (X = H) and ionic forms (X = Mg) of MCPA are included. Interaction energies were estimated using quantum chemical calculations to allow a relative comparison. Interaction energies in the range of ±12 to ±112 kJ mol ±1 were calculated for charge transfer and several hydrogen bond interactions which are likely to occur in sorption of polar molecules (such as MCPA) on organic matter (Dec & Bollag, 1997). The interaction energies obtained are in good agreement with published data, where similar systems were investigated (Leach, 1996;Aquino et al., 2000;Tunega et al., 2000). While a value of about ±12 kJ mol ±1 is in the range of weak van der Waals interactions, the other values are typical for polar dipole interactions (Leach, 1996;Kubicki & Apitz, 1999). Electrostatic interactions with silicate, aluminate or organic matter showed distinctly larger interaction energies (up to ±329 kJ mol ±1 ). If such electrostatic sites are accessible for MCPA, they are likely to be involved in the sorption of MCPA (McBride, 1989;Cox et al., 1997). Due to the large interaction energy, desorption of molecules bound to such adsorption sites seems less probable (Burgos et al., 1996;Pignatello & Xing, 1996), and sorption of MCPA may be less reversible (within the experimental time span) the more such binding sites are present.
The signi®cantly smaller adsorption±desorption quotients (Table 3) from samples from the fallowed soil in comparison with ones from the other treatments may be partly due to their having more polar and electrostatic binding sites. Since the fallowed soil contains the least organic matter the enhanced contribution of inorganic compounds (Murphy et al., 1990;Celis et al., 1997;Shen, 1999) could be one factor explaining the differences in adsorption±desorption behaviour. However, since other factors, such as changes in entropy, hydration enthalpy or in diffusion processes, are not available from such simple interaction models, we cannot obtain any quantitative estimates of the contribution of each of the factors to the overall sorption behaviour.

Conclusions
Different farming practices over a long time changed the adsorption±desorption properties of soil for MCPA. The origin of organic matter seems to play an important role in the sorption, which was signi®cantly altered by the different amendments. Sewage sludge had the largest effect. The other treatments were peat amendment, animal manure, green manure, inorganic fertilizers without N, and fallow, in that order. The increased sorption of MCPA on the soil receiving peat and sewage sludge can be related both to an increase in carbon content and to a decrease in pH relative to the other treatments. However, besides the content the origin and characteristics of the soil organic matter seem to be crucial for the sorption behaviour. Adsorption of MCPA was less on the peat-amended soil than on the soil treated with sewage sludge, even though the former contained more organic carbon. Additionally, the fallowed soil, which contains the least organic carbon, adsorbed MCPA similarly to the soil that received green manure and animal manure. Both the different origin of soil organic matter and an increasing role of mineral sorption with decreasing carbon content could contribute to this behaviour.