Pest Management Science

Pest Manag Sci 60:417433 (online: 2004) DOI: 10.1002/ps.840

Transport and distribution of lindane and simazine in a riverine environment: measurements in bed sediments and modelling
Ian J Allan,1,2 William A House,1 Andrew Parker2 and Joy E Carter3
1 Centre

for Ecology and Hydrology, Winfrith Technology Centre, Dorchester, Dorset, DT2 8ZD, UK Institute for Sedimentology, University of Reading, Whiteknights, Reading, Berkshire, RG6 2AB, UK 3 University of Glamorgan, Pontypridd, CF37 1DL, UK
2 Post-graduate

Abstract: Aquatic sediments often remove hydrophobic contaminants from fresh waters. The subsequent distribution and concentration of contaminants in bed sediments determines their effect on benthic organisms and the risk of re-entry into the water and/or leaching to groundwater. This study examines the transport of simazine and lindane in aquatic bed sediments with the aim of understanding the processes that determine their depth distribution. Experiments in ume channels (water ow of 10 cm s1 ) determined the persistence of the compounds in the absence of sediment with (a) de-ionised water and (b) a solution that had been in contact with river sediment. In further experiments with river bed sediments in light and dark conditions, measurements were made of the concentration of the compounds in the overlying water and the development of bacterial/algal biolms and bioturbation activity. At the end of the experiments, concentrations in sediments and associated pore waters were determined in sections of the sediment at 1 mm resolution down to 5 mm and then at 10 mm resolution to 50 mm depth and these distributions analysed using a sorptiondiffusiondegradation model. The ne resolution in the depth prole permitted the detection of a maximum in the concentration of the compounds in the pore water near the surface, whereas concentrations in the sediment increased to a maximum at the surface itself. Experimental distribution coefcients determined from the pore water and sediment concentrations indicated a gradient with depth that was partly explained by an increase in organic matter content and specic surface area of the solids near the interface. The modelling showed that degradation of lindane within the sediment was necessary to explain the concentration proles, with the optimum agreement between the measured and theoretical proles obtained with differential degradation in the oxic and anoxic zones. The compounds penetrated to a depth of 4050 mm over a period of 42 days. 2004 Society of Chemical Industry

Keywords: simazine; lindane; algal biolm; river sediment; pesticide degradation; pesticide diffusion in sediment

1 INTRODUCTION River bed-sediment is an important sink for hydrophobic pesticides and other micro organic contaminants (MOCs). The partitioning of MOCs between river compartments1 and their transport2 and degradation3,4 are decisive processes in the evaluation of the risk to river fauna, and possible threats to coastal and ocean ecosystems and ground waters. The kinetics and extent of the association of MOCs with sediment has been the subject of few studies.5 Furthermore, knowledge of their transport6 by diffusion within bed-sediments is limited.7 9 The present study investigates the diffusive transport of lindane and simazine in bed sediment from the River Calder, using experimental umes to mimic lotic fresh waters.10,11

The ow over the sediment was chosen to be low enough (10 cm s1 ) to avoid suspension of sediment and to permit only diffusive transport, with minimal advective ow at the sedimentwater interface. The movement of MOCs exclusively through diffusion in pore waters is inuenced by: (1) sorption to sediment particles/colloids and their components, eg organic matter and clay; (2) sorption to, and metabolism at, the sedimentwater interface by algal and bacterial biolm; (3) degradation through biotic/abiotic processes; and (4) bioturbation by macro-organisms, eg oligochaete worms, through physical mixing. The afnity of MOCs to sediment has been shown to have an effect on their depth penetration in river sediments,12 and of lindane

Correspondence to: Dr William A House, 30 Filleul Road, Sandford Woods, Wareham, BH20 7AP, UK E-mail: williamalanhouse@aol.com Contract/grant sponsor: National Environmental Research Council (CASE award) (Received 10 February 2003; revised version received 2 July 2003; accepted 7 November 2003) Published online 6 February 2004 417

2004 Society of Chemical Industry. Pest Manag Sci 1526498X/2004/$30.00

IJ Allan et al

and simazine in mesocosm experiments.13 These processes may be incorporated in models to predict the distribution, persistence and movement of MOCs. The objectives of this study were: (1) to determine the vertical depth distribution of lindane and simazine in a 50-mm-deep bed of sediment with a ne-resolution (ca 1 mm) at the sedimentwater interface, both in the whole sediment (ie sedimentbound plus pore water), and in pore waters; (2) to examine the effects of the development of an algal and bacterial biolm at the sedimentwater interface, and (3) to compare observed concentration proles with those generated by a one-dimensional sorptiondiffusiondegradation model. The work was conducted in three steps: (a) an initial study of the fate of the compounds in the absence of sediment, (b) experiments involving a sediment-bed in dark/natural light conditions to examine the effect of an algal biolm, and (c) modelling the concentrationdepth proles to account for sorption and degradation processes in the sediment.

AIR SUPPLY SEDIMENT BED FRIT

OVERLYING WATER

COOLING TANK WITH WATER FROM RIVER FROME VALVE AND TAP FOR OVERFLYING WATER SAMPLING

FLOW TRANSDUCER CENTRIFUGAL PUMP

Figure 1. Diagram of an experimental uvarium channel.

2 EXPERIMENTAL 2.1 Chemicals Simazine (99.9% purity) was supplied by the laboratory of Dr Ehrenstorfer (D-86199 Augsburg, Germany), and lindane (99.9% purity) was obtained from The Laboratory of the Government Chemist (Teddington, UK). The internal standard ametryn was provided by Promochem/Riedelde Haen (D3016 Seelze 1, Germany). Pesticide grade acetone, methanol and ethyl acetate, AR grade anhydrous sodium sulfate and potassium hydrogen carbonate were purchased from BDH (Poole, UK). Milli-Q water was provided by an Elix water purication system gradient A10 (Millipore). Stock solutions of simazine and lindane (200 mg litre1 in ethyl acetate) were used for analytical standards and for spiking solutions for ume experiments. 2.2 River sediment Surface bed-sediment was collected from the River Calder (Humber catchment) in the North-East of England at Methley Bridge (NGR SE409258) in May 2000. This was a contaminated site that had been studied previously.14 Sediments from a depth <5 cm were gathered using a stainless steel scoop, sieved onsite (2-mm stainless steel) and transported in 10-litre plastic cylinders to the laboratory, where the overlying water was continuously aerated. The sediments were thoroughly mixed before being introduced into duplicate umes. 2.3 Experimental umes Flumes were initially developed to study nutrient dynamics in streams.11,15,16 The 2-m long, 10-cmwide acrylic channels are designed to bring into contact a 50-mm-deep bed of sediment with aerated overlying water with control of the water velocity
418

(Fig 1). The water is re-circulated using a magnetically coupled centrifugal pump, with the ow measured by a transducer and controlled by a buttery valve. The preparation of a at bed of sediment and the use of low ow rates avoided sediment re-suspension, and permitted the study of transport exclusively through diffusion in pore waters. The umes were housed in a uvarium with a glass roof and immersed in a tank of owing water from the River Frome. 2.4 Experiments without sediment The channels were used in two experiments without a sediment-bed and containing approximately 24 litres of 10 mM potassium hydrogen carbonate (same ionic strength as the river water). Variations of pH, dissolved oxygen, electrical conductivity and temperature were measured at different times of the day over 15 days using eld meters (Ciba-Corning M90) previously calibrated using standard procedures. Pesticide solutions were evaporated to near dryness under nitrogen and reconstituted in acetone to obtain 2 ml spiking solutions containing ca 2.4 mg each of simazine and lindane. At the start of the experiment, the solutions were spiked and samples collected for analysis. De-ionised water was added to keep the water volume constant. During the two experiments, Channel 1 was in the light while Channel 2 was in the dark. The solution in the second experiment was previously in contact with river sediment from the River Calder for a period of six weeks. 2.5 Experiments with a sediment-bed in light/dark conditions Two experiments of six weeks duration were conducted. In the rst of these the channels were in the dark, and in the second they were in the light, permitting the growth of a photosynthetic biolm. A at sediment-bed was laid down in the channels and 20 litres of 10 mM potassium hydrogen carbonate were added to each channel, care being taken not to disturb the bed. The sediments were left for 6 days before spiking with a pesticide mixture containing 10 mg each of simazine and lindane in 2 ml of acetone prepared from the bulk solutions as described in Section 2.4. Populations of oligochaete
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

worms were left to develop naturally. Surface sediments (ca 1 mm depth) were collected using a spatula with a knife-edge, and development of the biolm was monitored weekly during the experiment under light conditions. The penetration of dissolved oxygen into the sediment was measured using microelectrodes as described previously.11 At the end of the experiments, after removing the overlying water, the sediment was longitudinally sectioned (using a pre-calibrated slicing tool) at 1-mm intervals down to 5 mm below the sedimentwater interface, then 5 to 10 mm, and then every 10 mm down to 50 mm. In the case of experiments under light conditions, the algal biolm was sampled separately. Sediments were subsequently sub-sampled for pesticide analysis and characterisation. The porosity and organic matter content of the sediment were determined by drying a known volume and mass at 105 C for 15 h and subsequently at 550 C for 5 h. After centrifugation, pore waters were analysed for dissolved silicon by ow injection analysis using a FIAstar 5012 spectrophotometer and analyser. The measurement allows some estimate of the importance of bioturbation.10 Algal biolm and sediment characterisation was undertaken by extracting chlorophyll a (0.7 g solids) with methanol (8.0 ml) vigorously shaken for 1 h, and analysed using a Beckman DU-520 spectrophotometer with a 10-mm cell at 665 and 750 nm. Extra-polymeric substances (EPS) secreted by micro-organisms during metabolism or algal growth were estimated by measuring total carbohydrates by phenolsulfuric assay.17,18 The cation-exchange capacity was also determined.19 Whole-rock and clay mineralogy was assessed by X-ray diffraction with a Philips PW1710 goniometer. Surface area was determined using a Coulter Surface Area 3100 instrument and nitrogen gas, and analysed using the BET equation.20 2.6 Extraction of simazine and lindane 2.6.1 Aqueous samples Solid-phase extraction (SPE) was used with an Isolute 6 ml silica-bonded C18 (CE) cartridge containing sorbent (1 g) tted onto a Vac-Elut (Baker SPE10) unit, itself connected to a bench-size KNF (Neuberger) pump. The cartridge was conditioned with methanol (6 ml) for 5 min, then replaced by 10 ml of Milli-Q water, followed by 300 ml of sample. The sample bottle was rinsed with ethyl acetate and the compounds eluted with ethyl acetate (6 ml). Extracts were dried and reduced under nitrogen. 2.6.2 Pore water samples Sediment sections were centrifuged with an MSE 18 high-speed centrifuge for 60 min at 6000 rev min1 . To minimise adsorption of lindane and simazine on hardware surfaces, solvent-rinsed stainless steel tubes were used. Interstitial waters were subsequently extracted by SPE as described above.
Pest Manag Sci 60:417433 (online: 2004)

2.6.3 Sediment samples Samples were freeze-dried (Edwards Modulyo) prior to supercritical uid extraction (SFE) using a Dionex SFE-703 system with a co-solvent addition module using 80% SFC/SFE grade carbon dioxide and 20% methanol. The extraction was done at an oven temperature of 50 C and a cell pressure of 400 atm for 60 min. Flow restrictors were heated to 100 C. Collection vials containing ca 15 ml of ethyl acetate were kept at 4 C and volume of solution subsequently reduced to 2 ml under nitrogen before drying with anhydrous sodium sulfate. 2.7 Gas chromatography/mass spectrometry analysis (GC/MS) One-millilitre samples in ethyl acetate were spiked with an internal standard using 10 l of ametryn (50 g ml1 ) prior to analysis by GC/MS. The analysis was performed using a Perkin-Elmer Autosystem XL GC coupled to a Turbomass mass spectrometer. Samples (5 l) were automatically injected with a split/splitless system and separated on a capillary column (30 m long, 250 m diameter, 0.25 m lm thickness) with a DB-5 phase (J&W Scientic), with helium as carrier gas (1.0 ml min1 ). The oven programme consisted of an initial temperature of 100 C for 2 min, a rst step at 170 C with a rate of 30 C min1 , and a second step a 280 C with a rate of 5 C min1 . The injector temperature was initially set to 100 C followed by an increase at the rate of 200 C min1 until 285 C was attained, and then held. The detection was in positive ionisation (+EI at +70 eV) with single-ion recording after the main and conrmation ions were selected in full scan (m/z [50500]). Main and conrmation ions were 173 and 201 for simazine, and 109 and 181 for lindane. Recoveries for the SPE were 110 (4)% and 86 (8)% and for SFE of 94 (17)% and 80 (10)% for simazine and lindane respectively.

3 MODELLING METHODOLOGY 3.1 Theory of the modelling of vertical concentration proles in the bed-sediment Diffusion in the pore water in a one-dimensional system, where the concentration in the overlying water is homogeneous above a uniform bed, may be expressed as:21 CMOC 2 CMOC = DSed t x2 (1)

where CMOC is the MOC concentration (g m3 ), x the depth (m), DSed the sediment diffusion coefcient (m2 s1 ) and t is time (s). Diffusion in sediment can be related to the molecular diffusion22 through the Weissberg relation22 with the dimensionless tortuosity ( ): DSed = Dm Dm = 2 1 Ln( 2 ) (2)
419

IJ Allan et al

where Dm is the molecular diffusion coefcient at innite dilution (m2 s1 ) and is the porosity. The interaction between sediment particles and MOCs may be expressed in terms of reversible sorption. For low concentrations of solute, a constant distribution coefcient Kd describes the linear part of the sorption isotherm: Kd = Csorbed /Cwater (3)

WilkeChang may be used for the determination.24 In particular, the WilkeChang relationship, accurate for large molecules in low-viscosity uids, has been used by many researchers for the theoretical determination of diffusivities of aromatics in water, or to compare with experimental values.24 27 Dm = 7.4 108 (xW MW )1/2 T 0.6 W VBMOC (5)

where Csorbed and Cwater are the concentration of MOC sorbed to sediment and dissolved in pore water respectively. This relationship is true for sufciently dilute systems with fugacity coefcients5 approaching limiting values. Incorporating eqns (1)(3) and including degradation by rst-order kinetics (described by a rate constant, kdeg ): CMOC Dm = t (1 + Kd )(1 ln( 2 )) 2 CMOC kdeg CMOC x2 (4)

where T is the absolute temperature, xw the association number of the solvent (2.6 for water22 ), MW the molecular mass of water, W the viscosity of water (cP) and VBMOC the molar volume of the solute at its boiling point. The incremental method of LeBas28 30 for the theoretical determination of the molar volume, and an empirical formula22 for the calculation of the viscosity of water were used. 3.2.2 Changes in porosity with depth in the sediment-bed Changes in porosity with depth are a result of compaction, experimental conditions and sediment characteristics. Logarithmic, linear and power laws have been successfully developed to describe porosity proles in sediment.31,32 The representation of porosity proles with depth by a power law relationship gave excellent agreement (see Fig 2) for the present datasets (r 2 > 0.89), resulting in similar coefcients for the power equation: %porosity = 68.9 (depth)0.123 (6)

Degradation is characterised as a process incorporating both chemical and biological breakdown. This approach is approximate, as the degradation rate constant, kdeg , may vary with time or depth in the sediment. For biotic degradation, time dependency is often caused by microbial adaptation to the MOC, and for degradation in oxic/anoxicanaerobic conditions the rate constant may vary with depth. A program in FORTRAN, previously written and developed for modelling silica dissolution and uxes in freshwater sediments,10,23 was transformed for a stepwise resolution of the system of differential equations by numerical integration. The model generates concentration gradients in the whole sediment and pore water along the depth in the sediment-bed. It uses the concentrations in the overlying water, determined experimentally, as the boundary conditions, ie the concentrations in the sediment layers are predicted from the observed changes in the concentration in solution. The predicted pore water prole is obtained initially, while whole sediment concentrations are calculated for each sediment section using the particle density, porosity and Kd values. The equation has been extended to include the effects of different degradation rates in the oxic and anoxic compartments of the sediment and assimilation of MOCs in a biolm as well as irreversible sorption to the sediment. 3.2 Procedure for the modelling the concentration-depth proles 3.2.1 Theoretical determination of the molecular diffusion coefcients in water A decisive input to the model is the diffusivity of the pesticide at innite dilution. There are no published data available for lindane and simazine. Empirical equations such as the StokesEinstein or
420

where the depth is in mm. This is incorporated in the model to account for changes in porosity with depth. The application of this relation to the duration of the experiment is supported by other work7 which showed no signicant change in porosity or organic matter content with time (ie 70 days) for different sediment layers. 3.2.3 General method for optimisations in the algorithm Optimisations are based on the root mean square deviation between predicted and observed wholesediment concentration-depth proles, RMS. A coefcient, RMSCOEF, is employed to give more weight to values far from the limit of detection (LOD) and to lower the impact of those closer to LOD values to reect the error of the chemical analysis. For the highest experimental value, Cmax , RMSCOEF = 1, while for an experimental value at the LOD, the coefcient is set to 0.5 and varied linearly between these two values. 3.2.4 Optimisation of an organic matter-normalised distribution coefcient, KOM The assumption here is that the MOC distribution is controlled by a constant organic matter-normalised distribution coefcient, KOM given by KOM = Kd /fOM
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

where fOM is the fraction of organic matter. This is independent of time or depth in the sediment. Equation (4) may be transformed by substituting for Kd in terms of KOM and fOM . 3.2.5 Input of experimentally determined Kd values for each sediment section It is assumed that Kd values calculated from experimental pore water and whole sediment concentrations (eqn (3)) for each sediment layer are true measures of the instantaneous Kd values that occurred during the experiment. In this case no optimisation is necessary. 3.2.6 Optimisation of degradation rates Degradation/transformation processes are modelled by applying a rst-order degradation rate constant, independent of the depth in the sediment. Sorption becomes an input to the model using either the optimised KOM or experimental Kd values for each section. The degradation constant, kdeg , is optimised using the procedure described above. Aeration of the overlying water, oxygen production by photosynthetic activity at the sediment surface, and hence oxygen penetration, creates two distinct compartments exhibiting different chemical environments, eg pH and redox conditions, and microbial communities. It has been shown that degradation differs in terms of pathways and rates in oxic or anoxic/anaerobic conditions.33 35 Therefore the use of distinct rstorder degradation rates for oxic and anoxic zones may be appropriate, and involved the optimisation of two variables, kdeg1 and kdeg2 , for the oxic and anoxic zone respectively, using a SIMPLEX optimisation.36

and lindane to the edges, or degradation in solution over the two weeks. Changes in the concentration of simazine and lindane could be accounted for by the dilution effect of the sampling and subsequent replacement of water. However, the second experiment, using water that had been in contact with River Calder sediment for a six-week period prior to the experiment, showed losses of simazine and lindane from the bulk solution with simazine decreasing to <200 g litre1 and lindane to <0.01 g litre1 . 4.2 Biolm growth and sediment-bed characteristics The porosity proles in Channels 1 and 2 under dark and light conditions are represented by a power law (Fig 2a,b). The amount of organic matter in the prole increased towards the sediment surface (Fig 2c,d). Sediments from the experiment in the light generally showed higher specic surface areas compared with sediments from those in the dark for the deeper layers (Fig 2e,f), and an increase close to the surface. As expected, the measurement of chlorophyll a illustrates the growth of the algal biolm at the surface under light conditions, with values reaching 35 g g1 dry weight of sediment for the top layer (Fig 2h). However, under dark conditions, a peak of chlorophyll a is also present (Fig 2g). There is an increase in the extra polymeric substances content in top layers under dark conditions, while more variability is shown by proles obtained in the light. The trend for proles under light conditions is similar to those in the dark, with a similar range of values reached at depth and at the sediment surface. The oxygen microelectrode measurements indicated a depth of oxygen penetration of approximately 23 mm below the sedimentwater interface. The analysis of bulk sediment particles showed that quartz was the main mineral (73% of total minerals) and clays were all under 5%. The samples also contained calcite, albite, microcline, kaolinite and haematite. The cation-exchange capacity (CEC) was determined for samples from three sections. Results showed slightly higher CEC for the 12-mm sections than for the 1020-mm sections with the CEC varying between 1.6 to 9.04 mequiv 100 g1 sediment. The CEC values determined for the 4050-mm sections

4 EXPERIMENTAL RESULTS 4.1 Persistence and sorption in experiments without sediment In the bulk solution during the two experiments without sediment, the temperature was higher during the experiment with de-ionised water and the pH was lower when using river water (Table 1). Dissolved oxygen remained over 90% and the conductivities were in similar ranges. The rst experiment with the channels containing de-ionised water revealed no sorption of simazine

Table 1. Mean values of temperature, conductivity, pH and dissolved oxygen (DO) in the bulk solution for all experimentsa

Experimental specications No sediment, de-ionised water No sediment, river water With bed-sediment Light Dark Light Dark Dark Light
a

Temperature ( C) 11.6 (10.812.6) 14.2 (13.914.7) 15.3 (12.817.2) 14.9 (10.418.2)

Conductivity (S cm1 at 25 C) 1172 (10401221) 976 (9211012) 821 (802943) 720 (702756) 1246 (9751503) 1338 (10341580) 1146 (10091407) 1073 (10261464)

pH 8.8 (8.629.11) 7.8 (7.68.1) 7.8 (7.18.8) 8.4 (7.79.1)

DO (%) 99 (98100) 95 (9398) 93 (9198) 94 (90100) 92 (8797) 92 (8796) 99 (86102) 99 (93103)

The ranges of the parameters are shown in brackets.

Pest Manag Sci 60:417433 (online: 2004)

421

IJ Allan et al
Porosity (%) 40 0 Depth (mm) 10 20 30 40 50 (a) Dark Organic Matter (%) 5 6 Ch. 1 Ch. 2 power equation 50 60 70 80 0 10 20 30 40 50 (b) Light Organic Matter (%) 5 6 Ch. 1 Ch. 2 power equation 40 50 Porosity (%) 60 70 80

3 0 Depth (mm) 10 20 30 40 50

8 0 10 20 Ch. 1 Ch. 2 30 40 50

Ch. 1 Ch. 2 (d) Light Surface Area (m2 g1) 2.5 3 3.5

(c) Dark Surface Area (m2 g1) 2.5 3 3.5

1.5 0 Depth (mm) 10 20 30 40 50

1.5 0 10 20 30

Ch. 1 & 2 (e) Dark Chlorophyll a (g g1 dry sediment) 10 20 30

40 50 (f) Light

Ch. 1 & 2

0 0 Depth (mm) 4 8 12 16

40 0 4 8 Ch. 1 Ch. 2 12 16

Chlorophyll a (g g1 dry sediment) 10 20 30

40

Ch. 1 Ch. 2 (h) Light EPS Concentration (mg g1 dry sediment)

(g) Dark EPS Concentration (mg g1 dry sediment) 0 10 20 30 40

0 0 10 20 30

10

20

30

40

0 Depth (mm) 10 20 30 40 50 (i) Dark Ch. 1 Ch. 2

40 50 (j) Light

Ch. 1 Ch. 2

Figure 2. Proles of certain sediment characteristics in the sediment-bed of experiments under dark and natural light conditions, including (a and b) porosity, (c and d) organic matter, (e and f) surface area, (g and h) chlorophyll a and (i and j) extra-polymeric substances (EPS).

were between 11.5 and 19.8 mequiv 100 g1 sediment and slightly higher than for the upper sections. Oligochaete worms naturally present in the sediment showed decreasing activity after addition of the pesticides. The analysis of dissolved silicon10 proles
422

(not presented here) showed no sign of sediment pore water reworking. The densities measured at the end of the experiment (700900 individuals m2 ) were considerably lower than in specic bioturbation studies.37
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

4.3 Distribution of MOCs in sediment and temporal changes in bulk solution The conditions for the two experiments are summarised in Tables 1 and 2. With the exception of potassium, all ions listed in Table 2 were not initially added to the system but are the result of diffusion from the sediment. Sodium, potassium, uoride, chloride and nitrate ions were present at lower concentrations during the experiment under natural light conditions.

4.3.1 Simazine concentrations The variation in concentrations in the overlying water with time, and whole sediment and pore water concentration-depth proles at the end of the experiments, are shown in Fig 3. In the dark, the dissolved simazine concentration decreased more rapidly in Channel 2 than in Channel 1, resulting in consistently lower whole sediment and pore water concentrations in Channel 2. Whole sediment concentrations were

Table 2. Concentrations of the major ions at the end of the experiment under light conditions in Channels 1 and 2a

Concentration (mg litre1 ) Ions Channel 1 Channel 2

a

F (0.21) (0.22)

Cl 4.7 (16.0) 3.9 (17.4)

NO3 -N 1.5 (10.7) 1.3 (10.7)

SO4 2 -S 221 (89) 227 (105)

Na+ 5.8 (22.0) 6.3 (24.7)

K+ 8.2 (124) 7.5 (126)

Mg2+ 13.8 (12.9) 14.1 (13.4)

Ca2+ 87.2 (56.6) 88.5 (62.4)

In brackets are the values determined for the experiment in the dark.

Dark Conditions

Light Conditions

Concentration (g litre1)

140 120 100 80 60 40 20 0 0 5 10

(a) Overlying water

200 150 100 50 0 0 5 10

(d) Overlying water

15

20

25

30

35

40

15

20

25

30

35

40

Time (days) Concentration (g kg1 dry wt of sediment) 0 10 Depth (mm) 20 30 40 50 (b) Whole sediment profiles Concentration (g litre1) 0 10 Depth (mm) 20 30 40 50 (c) Pore water profiles 0 2 4 6 8 10 12 14 16 0 0 10 20 30 40 50 5 0 50 100 150 200 0 10 20 30 40 50 0

Time (days) Concentration (g kg1 dry wt of sediment) 50 100 150 200

(e) Whole sediment profiles Concentration (g litre1) 10 15 20 25

30

35

(f) Pore water profiles

Figure 3. Results for simazine under dark and light conditions. (a,d) temporal changes in concentration of simazine in the overlying solution of Channels 1 (open symbol) and 2 (lled symbol) and resulting (b,e) whole sediment and (c,f) pore water concentration-depth-proles at the end of the experiment.

Pest Manag Sci 60:417433 (online: 2004)

423

IJ Allan et al

generally higher towards the sedimentwater interface, while pore water proles exhibited peaks at 5 mm depth (Fig 3). After 1520 days, the overlying water concentrations dropped from 140 g litre1 to 1 g litre1 and remained low until the end of the experiment. Simazine penetrated to a depth of 4050 mm in Channel 1, though it was only detected to a depth of 30 mm in Channel 2. A maximum pore water concentration of 14 g litre1 was observed in Channel 1 in the 45-mm section and the nal pore water concentrations were found to be higher than overlying water concentrations. In comparison, overlying water concentrations in the experiment in the light (Fig 3) were high (80140 g litre1 ) for the rst 510 days, followed by a sharp drop to <1 g litre1 until the end of the experiment. Whole sediment concentration-depth proles were similar to those obtained in the dark, with the highest values for the top section, and penetration down to the 4050 mm depth. Pore water concentration-depth proles were in good
Dark Conditions Concentration (g litre1) 500 400 (a) Overlying water

agreement with whole sediment and overlying water concentrations, since concentrations were generally higher for Channel 1 than for Channel 2. Pore water concentrations were higher than the overlying water concentrations at the end of the experiment. Although whole sediment proles in dark and light conditions were similar, pore water concentrations were higher under light conditions. At the end of the experiment, measured concentrations in the biolm were 53 and 7 g kg1 (dry weight) for Channels 1 and 2 respectively, giving rise to calculated distribution coefcients between biolm and overlying water of 100 and 700 litre kg1 at the end of the experiment. 4.3.2 Lindane concentrations Temporal changes in the overlying water concentrations during the experiment under dark conditions showed maxima of 436 and 117 g litre1 for Channels 1 and 2 respectively, 412.5 h after spiking (Fig 4). Concentrations decreased to 1 g litre1 before midexperiment and remained low until the end. Similarly,
Light Conditions (d) Overlying water 800 600

300 200 100 0 0 5 10 15 20 25 Time (days) 30 35 40 400 200 0

10

15

20

25

30

35

40

Time (days) Concentration (g kg1 dry wt of sediment) 200 400 600 800

Concentration (g kg1 dry wt of sediment) 0 10 Depth (mm) 20 30 40 50 (b) Whole sediment profiles Concentration (g litre1) 2 3 4 5 6 0 50 100 150 200 250 0 0 10 20 30 40 50

(e) Whole sediment profiles Concentration (g litre1) 1 2 3 4

0 0 10 Depth (mm) 20 30 40 50

8 0 10 20 30 40

(c) Pore water profiles

50

(f) Pore water profiles

Figure 4. Results for lindane under dark and light conditions: (a,d) temporal changes in concentration of lindane in the overlying solution of Channels 1 (open symbol) and 2 (lled symbol) and resulting (b,e) whole sediment and (c,f) pore water concentration-depth-proles at the end of the experiment.

424

Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

wide differences in concentrations were observed for both channels in the light (Fig 4), while concentrations reached values below 0.5 g litre1 at the middle of the experiment. Whole sediment proles exhibited concentration gradients with the highest concentrations at the top, while concentrations deeper in the sediment were close to detection limits. Concentrations obtained in the light are higher than those obtained in the dark, a result not reected in pore water concentration-depth proles (Fig 4). All pore water concentration proles showed similar trends, with a peak in the 45-mm section in light conditions and deeper in dark conditions at 1020 mm. High concentrations were observed in the algal biolm at the end of the experiment under light conditions, with concentrations of 1073 and 884 g kg1 (dry weight) for Channels 1 and 2 respectively. Calculated distribution coefcients between biolm and water were 4300 and 6800 litre kg1 at the end of the experiment. 4.3.3 Temporal changes in surface sediment/biolm A decrease in concentration was observed for simazine and lindane between 11 and 31 days after exposure (Table 3). Chlorophyll a contents generally increased with time, while EPS contents were variable. The organic matter content was generally higher than the value obtained at the end of the experiment from sectioning the whole channel for the 01-mm section (Fig 2d). Porosity, similarly to EPS, showed a wide variability, with values in the range 4995%.

Mean values of 4.94 (0.06) and 4.75 (0.06) 1010 m2 s1 were obtained for simazine and lindane respectively.

5 MODELLING OF THE CONCENTRATION-DEPTH PROFILES The LeBas method gave values for the molecular volumes of 228.4 and 243.6 cm3 mol1 for simazine and lindane respectively. A molecular diffusion coefcient was calculated for each compound for each experiment. However, the very similar temperature conditions during both experiments (Table 1) meant that a single value could be used for each compound.

5.1 EFFECTS OF REVERSIBLE SORPTION The use of single Kd values to describe reversible sorption was not able to predict the large changes in the whole sediment proles found experimentally for either compound. The use of a constant KOM value together with the knowledge of the variation in fOM (Fig 2c,d) was included in eqn (4) with kdeg = 0. Optimised KOM values are shown in Table 4. Root mean square deviation values were in the range 0.3652.3. Examples of the comparison of the proles are given in Fig 5. Distribution coefcients were calculated from experimentally determined whole sediment and pore water concentrations for each sediment section (Table 5). For simazine in dark conditions, Kd values were very variable, with highest near the sediment surface. The wide variability for deeper layers may be the result of whole sediment concentrations close to limits of detection. For simazine in the experiment in light conditions, Kd values ranged from 24 litre kg1 for the top layer, decreasing to values close to or below one. For lindane in dark conditions, Kd varied from 439 and 289 litre kg1 for Channels 1 and 2 respectively for the 01-mm sections, down to values of 9.5 and 22 for the 2030-mm section. Calculated Kd values for the light experiment indicated a similar decrease, with maximum concentrations observed for the 01-mm section below the sedimentwater interface. However, Kd values were higher than those found for the experiment under dark conditions. Optimising KOM gave values of between 70 and 366 litre kg1 (mean 164 litre kg1 ) for simazine and 11 and 999 litre kg1 for lindane (Table 4). The results were variable for simazine, with no signicant difference in the KOM values obtained in light and dark conditions. In contrast, the results for lindane indicated higher sorption in light conditions, as

Table 3. Temporal changes in simazine and lindane concentrations in the top 1-mm section of sediment/biolm during the experiment in light conditions; values porosity and chlorophyll a (Ch 1), extra polymeric substances (EPS) and organic matter (OM) contents for sediment samples are also given

Concentration (g kg1 DW)a (SEM) Time after spiking (days) Channel 1 11 17 24 31 Channel 2 11 17 24 31
a

Simazine 560 (96) 104 (18) 164 (28) 70 (12) 154 (27) 38 (7) 99 (17) 26 (5)

Lindane 3556 (363) 624 (64) 563 (58) 417 (43) 7303 (745) 692 (71) 693 (71) 237 (25)

Chl a (g g1 ) 4.7 41.1 27.9 40.5 5.1 47.6 67.5 85.1

EPS (mg g1 ) 13.3 18.5 39.4 13.1 18.1 15.8 22.2 6.6

OM (%) 10.0 9.9 7.0 12.6 8.3 6.7 9.9 17.3

Porosity (%) 78.3 90.7 95.9 83.9 79.9 82.8 76.8 49.7

DW = dry weight.

Pest Manag Sci 60:417433 (online: 2004)

425

IJ Allan et al
Table 4. Optimised values for the organic matter-normalised partition coefcient (KOM ), a single and two distinct rst-order degradation rate constants kdeg obtained from the sorptiondiffusiondegradation model

Two distinct kdeg values (107 s1 ) Optimised KOM value (litre kg1 ) Simazine Dark Light Lindane Dark Light
a

Single kdeg value (107 s1 ) 5.4 (14.9) 4.5 (17.8) 6.7 (11.9) 5.3 (15.1) 6.6 (12.1) 12 (6.7)

Oxic 16 (5.0) 14 (5.8) 13 (6.3) 8.5 (9.4) 13 (6.4)

Anoxic 2.8 (29.0) 0.7 (115.8) 0.3 (308.6) 9.5 106 0 2.2 (36.8)

Channel 1 Channel 2 Channel 1 Channel 2 Channel 1 Channel 2 Channel 1 Channel 2

366 94 125 70 23 12 999 356

In brackets are the corresponding calculated half-lives (days).

Table 5. Distribution coefcients determined at the end of experiments for each sediment layer and representation of changes in Kd with depth in the sediment-bed by a power law functiona

Experimentally determined Kd values (litre kg1 ) Simazine Dark Sediment sections (mm) 01 12 23 34 45 510 1020 2030 Channel 1 35 (445) 19 (291) 22 (444) 12 (262) 3 (67) 22 (503) 11 (234) 20 (428) Channel 2 72 (957) 52 (913) 24 (501) 6 (139) 6 (136) 38 (743) 61 (1246) 72 (1691) Channel 1 19 (256) 6 (112) 10 (191) 4 (83) 1 (21) 3 (61) 0.1 (2) 2 (39) 9.86 0.92 0.59 Light Channel 2 24 (308) 16 (237) 0.5 (10) 0.7 (14) 3 (62) 0.7 (14) 0.7 (14) 0.5 (10) 7.21 0.99 0.60 Channel 1 439 (5584) 125 (1917) 59 (1192) 47 (1025) 47 (1051) 8 (183) 6 (128) 10 (214) 181.0 1.14 0.90 Dark Channel 2 289 (3845) 145 (2547) 100 (2089) 33 (762) 14 (318) 5 (98) 6 (123) 22 (517) 130.4 0.99 0.68 Channel 1 748 (10 067) 252 (4702) 324 (6194) 86 (1783) 59 (1207) 94 (1909) 12 (230) 31 (609) 398.8 0.97 0.83 Lindane Light Channel 2 2624 (33 716) 1476 (21 821) 339 (6564) 156 (3156) 107 (2198) 125 (2518) 48 (970) 62

Modelling with Kd = A (Depth)b A b R2

a

1072.8 1.09 0.88

In brackets are the KOM values (litre kg1 ) calculated from the Kd values.

reected in the optimised KOM values (Table 4). The results for simazine obtained in the two experiments are comparable and illustrate that, although a peak in the pore water prole was observed, there was no corresponding peak in the whole sediment prole. The model, as expected, predicted that, given the low concentration of simazine in the overlying water at the end of the experiment (Fig 3), simazine would diffuse from the pore water back into the overlying water. As sorption is assumed reversible, this should result in a decrease in concentration in the whole
426

sediment near the sediment water interface. In both experiments the model predicted a peak in the whole sediment concentrations to match the peak in the pore water prole. As shown in Fig 5, the agreement between the theoretical and experimental proles is limited, although the pore water proles are of the correct shape and of the same order of magnitude as the measured values. The agreement was not much improved if the experimental Kd values were employed, (eg see Fig 5b). The low Kd values used in the model gave rise to a rapid movement of simazine in
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

Pore water concentration (g litre1) 0 10 20 30 0 0 Depth (mm) 4 (a) 8 12 16 0 10 10 20 30 Predicted Observed (b) 0 0 4 8 12 16 0 10 10 Predicted 1 Observed Predicted 2 (c) 20 30 40 50 (c) Predicted 1 Observed Predicted 2 40 50 0 50 (b) 100 150 200 Predicted Observed Predicted Observed 0 4 8 12 16 0 10 20 30 40 50 0 50 (a) 100 150 200 Predicted Observed Whole sediment concentration (g kg1 dry wt sed) 0 50 100 150 200

20

30

20

30

Figure 5. Comparison of measured and PW and WS concentration proles for simazine in Channel 1 of the experiment under dark conditions, after KOM optimisation of 366 litre kg1 (a), use of experimentally determined Kd values (b), optimisation of a single kdeg of 5.3 107 s1 (Predicted 1) and two distinct degradation rate constants kdeg1 and kdeg2 of 16.2 and 2.8 107 s1 , respectively (Predicted 2) (c).

the sediment and response to the drop in the overlying water concentration. Lindane also showed peaks in the pore water proles but not in the whole sediment proles, similarly to simazine. The concentration of lindane in the overlying water also decreased during the experiment (Fig 4), which is consistent with the gradient in the pore water prole, measured near the sedimentwater interface. The optimised KOM values computed in the two experiments were different (Table 4), reecting the higher concentrations in the whole sediment observed in light conditions. The model produced good agreement with the whole sediment prole to a depth of 5 mm for the experiment in the dark, but over-predicted whole sediment concentrations in deeper layers and over-predicted pore water concentrations through the prole. The use of Kd values calculated from the pore water and whole sediment proles produced better agreement between the pore water proles but considerably overestimated the concentrations in the whole sediment. The results using a power law to describe the variation of Kd with depth (Table 5) were similar to those using the individual values. In light conditions the predicted pore water prole was the correct shape and the concentrations were of the correct order of magnitude but, as with simazine, the predicted whole sediment
Pest Manag Sci 60:417433 (online: 2004)

concentrations showed a peak near the interface as a result of desorption as the pore water concentrations decreased (Fig 6). The agreement was not improved with the use of individual Kd values or a power function to describe the gradient in Kd (Fig 6b). 5.2 Effects of introducing degradation Degradation was introduced when the predicted proles generally overestimated the concentration proles in the sediment. This was not the situation for the prole for simazine in the light but applies to the results obtained in dark conditions. Improvement in the agreement with the experimental data was obtained with kdeg = 5.4 and 4.5 107 s1 for Channels 1 and 2, respectively, in the dark, corresponding to half-lives of approximately 15 and 18 days. In that case, while whole sediment proles were in excellent agreement, the model did not predict the peak of the pore water concentration for the 45-mm section (Fig 6c). The pore water and whole sediment proles were over-predicted for lindane if the individual Kd values were used (eg see Fig 6b). The optimisation of the degradation rates for lindane resulted in improvements in deviation between predicted and observed proles, with half-lives of 12 and 15 days determined for Channels 1 and 2 respectively in dark conditions. In light conditions, optimised half-lives of 12 and
427

IJ Allan et al
Pore water concentration (g litre1) 0 0 2 4 6 8 10 0 10 10 20 30 Predicted Observed (a) 0 0 2 4 6 8 10 12 14 0 10 Depth (mm) 10 20 Predicted 1 Observed Predicted 2 (b) 1 2 3 4 5 0 10 10 Predicted 1 20 Observed Predicted 2 30 0 0 1 2 (c) 3 4 5 0 10 10 20 20 Predicted 1 Predicted 2 Observed 30 (d) 30 40 50 (d) Predicted 1 Predicted 2 Observed 40 50 (c) 0 200 400 600 800 1000 20 30 Predicted 1 Observed Predicted 2 30 40 50 0 200 (b) 400 600 800 Predicted 1 Observed Predicted 2 40 50 0 1000 (a) 2000 3000 4000 Predicted Observed Whole sediment concentration (g kg1 dry wt sed) 0 200 400 600

20

30

20

30 0 0

Figure 6. Comparison of measured and predicted PW and WS concentration proles for lindane in Channel 1 of the experiment under natural light conditions, after KOM optimisation of 999 litre kg1 (a), use of experimentally determined Kd values (b; Predicted 1), use of the power-law equation (b; Predicted 2), optimisation of a single kdeg of 6.6 107 s1 (c; Predicted 1) and two distinct degradation rate constants kdeg1 and kdeg2 of 8.5 107 and 9.5 1013 s1 respectively (c, Predicted 2), and optimisation of KOM of 1100 litre kg1 and active uptake rate constant kass of 1.04 104 s1 (d; Predicted 1) or optimisation of a KOM value of 4006 litre kg1 and an irreversible rate constant kirr of 6.07 104 s1 (d, Predicted 2).

7 days were slightly lower than those found in dark conditions. Predicted proles obtained for lindane in dark and light conditions gave reasonable agreement with the shape and magnitude of the whole sediment and pore water proles (eg see Fig 6c). In most cases the agreement for the top 02-mm layers was less satisfactory, with the model overpredicting whole sediment concentrations in the top layers. Separating the bed into two compartments, an oxic layer and an anoxic, where degradation rates are likely to differ, may counteract this. In the case of simazine, the quality of the agreement for the data from Channel 1 in the experiment under dark conditions may be improved by optimisation of two distinct
428

rst-order degradation rate constants. As shown in Fig 5c, a lower degradation rate constant for the anoxic zone allowed slightly deeper penetration of simazine than with a single degradation rate. Halflives of 5.0 and 29 days were obtained for the oxic and anoxic layers respectively (Table 4). The optimisation for lindane in the dark experiment resulted in half-lives of 6 and 116, and 6 and 309, days for Channel 1 and 2, and oxic and anoxic compartments, respectively. In light conditions, half-lives of 9 and 6 days for the oxic layer, and zero degradation and 37 days, for the anoxic compartment for Channels 1 and 2 respectively (Table 4). In both cases the agreement was between the model prediction and observed data
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

was improved as shown by the lower root mean square deviation values. A higher degradation rate for the oxic layer allowed closer agreement for the whole sediment concentration of the 01-mm section, while the lower degradation rates for the anoxic compartment gave rise to deeper penetration in the sediment-bed (eg see Fig 6c). The use of two degradation constants for lindane provided an improvement in agreement with the PW proles.

sorption followed by degradation, may be responsible for the disappearance of compounds from solution. Studies in roto-torque reactors ushed with river water have demonstrated the sorption of micro organic contaminants such as lindane and atrazine to riverine bacterial biolm, and their subsequent degradation.1,4 The rst-order rate constants observed here for simazine and lindane (Table 6) are in agreement with the increase in sorption rate with increasing octanolwater partition coefcients observed by Headley et al 1 for a range of contaminants. 6.2 Whole-sediment and pore-water concentration proles Few studies have focused on the measurement of pore water concentrations and the establishment of high-resolution (ca 1 mm) pore water, whole sediment concentrations and Kd depth-proles in sediments. In a study of diffusion process of carbendazim in sediment,7 the sediment columns were divided into three layers for a depth of 7 cm, while other work9 used a 15-cm-long sediment column with the determination of three values of Kd along the core. In the present study, a comparison of the concentrations of the compounds in the 01-mm section at the end of the experiment in the light with those measured at different times during the experiment (Table 3), showed the importance of the measurement of a mean concentration for each sediment layer. The temporal changes indicated wide spatial variations, as shown by measurement of various sediment characteristics, and the heterogeneous nature of the sediment. The sectioning technique used here effectively averaged these variations over the sediment layer. Although concentrations in the overlying water differed from one channel to another, results for both channels in each experiment were in good agreement. Although whole sediment concentrations were signicantly higher in the experiment in light conditions, pore water proles exhibited higher maxima and higher concentrations between 5 and 25 mm below the sedimentwater interface. Pore water concentrations below the sedimentwater interface were very close for both experiments, despite signicantly different whole sediment concentrations. Therefore, it is believed that sorption of lindane does not follow a simple instantaneous reversible partition process. The low afnity of simazine for the sediment is illustrated by the depth-penetration of the compound to the bottom layers in 42 days, and by experimentally determined Kd values that are generally lower than for lindane (Table 5). Previous work with a different sediment showed a slightly higher diffusion of simazine13 caused by a lower porosity (5254%) and lower sorption to the sediment, probably because of its lower organic matter (OM) content. The greater penetration of simazine to depths of 15 cm after 45 days in simulated mudats3 is the result of diffusion
429

6 DISCUSSION 6.1 Sorption/degradation in the overlying water solution The loss of both compounds was caused by degradation in the solution or possibly a combination of sorption/degradation processes, although this was not observed using deionised water. As shown in Table 6, half-lives for simazine were higher than those for lindane, and both compounds disappeared faster in Channel 2 under dark conditions. The persistence of simazine in the experiment with de-ionised water is consistent with previous studies.38 Although the pH was slightly different in experiments with river water compared with de-ionised water, it was not low enough for hydrolysis to occur. At near-neutral pH, simazine is hydrolysed less than at high or low pH.39 An obvious positive effect of natural light radiations below 300 nm was observed on the hydrolysis/photolysis of simazine in water at pH 7.38 The glass roof of the uvarium stops natural UV radiation and hence may reduce potential photolytic reactions. The small differences in conductivity, temperature and dissolved oxygen are unlikely to be responsible for these losses. Important inuences on the composition of water previously in contact with river sediment are the occurrence of nutrients, organic matter and micro-organism communities initially present in the sediment. The presence of nutrients in solution permitted the growth and development of a micro-organism community in suspension or as a lm coated on surfaces. In these conditions, degradation, sorption, or a combination of
Table 6. Loss from bulk solution without sediment in the channels: rst-order rate constants and half-lives for simazine and lindane (and their standard errors), and half-lives and their ranges

Channel number

First-order-loss rate constant (107 s1 ) (SEM)

Half-life (days)

Experiment using de-ionised water Simazine 1: Light No signicant losses 2: Dark Lindane 1: Light 2: Dark Experiment using river water Simazine 1: Light 9.7 (0.4) 2: Dark 15.8 (0.7) Lindane 1: Light 38.3 (3.2) 2: Dark 51.3 (4.2) 8.3 (8.08.6) 5.0 (4.85.3) 2.1 (1.92.3) 1.6 (1.41.7)

Pest Manag Sci 60:417433 (online: 2004)

IJ Allan et al

and advective transport caused by the tides. Lindane, with a molecular diffusion coefcient close to that of simazine, also diffused to the bottom of the bed, though calculated Kd values were generally greater because of its hydrophobicity.30 Without degradation, much higher whole sediment concentrations may have been expected from the drop in concentrations in overlying water for both compounds. For simazine and lindane, the observed peak of pore water concentration and the concentration gradients decreasing towards the sedimentwater interface are a result of their release into the overlying water after their initial uptake. For periods over 20 days, the sediment-bed was in contact with low concentrations, which invoked a concentration gradient between the pore water and overlying water and subsequent backdiffusion. The increase in organic matter content in the top layers (Fig 2c and d), combined with changes in specic surface area, may have contributed to enhanced sorption in the sections just below the interface. Another possibility is the involvement of selective degradation rates in different compartments of the sediment. Since it has been shown that degradation rates of simazine in sediment may be inuenced by the presence/absence of oxygen,33 the existence of faster degradation rates in the oxic zone (ie depth <2 mm) may explain the lower pore water concentrations compared to deeper layers. Lastly, the amount of compound available for partition with pore water may be reduced because of the higher organic matter content of the surface layers. The presence of colloids in pore waters, not measured and accounted for in the present work, may have a major implication in the overall pore water concentrations. The effect may be relatively small for simazine but more important for lindane because of its greater hydrophobicity. The use of experimentally determined Kd values and power-law equations for lindane predicted more compound in the sediment than was observed. The compound in these cases was predicted to diffuse to less depth than observed. Diffusion is restricted by sorption and may indicate that the Kd was lower at the start of the experiment than the observed values at the end. This would allow the compound to remain for longer in pore water and subsequently diffuse deeper. This increase in sorption may be a result of the growth of the algal biolm, providing higher sorption at the sediment surface, or slow sorption/uptake mechanisms.6 It has been shown that 5% of lindane sorbed to sub-surface ne sand was due to slow sorption over a period of 167 days.6 Coefcients for the Kd versus depth power-law equation in the light were higher than those obtained for the experiment in the dark (Table 5). This may be the result of assimilation or bio-sorption to the biolm. This is supported by the high concentrations measured in the lamentous algal biolm at the end of the experiment. In addition, EPS (Fig 2i and j) secreted by micro-organisms may also affect the transport of lindane in particular.40
430

Measurements in the sediment-bed (eg EPS and chlorophyll a) and in the overlying water (eg dissolved oxygen or pH) clearly indicated the growth of an algal biolm at the sediment surface during the experiment under light conditions. However, monitoring of the experiment under dark conditions also demonstrated the presence of an active surface layer (Table 3). Both dark- or light-grown biolm may provide active sites for sorption1,40,41 with possible intramembrane diffusion, especially for simazine, designed to penetrate algal cells and membranes. Sorption to algae and micro-organisms may be followed by diffusion through cell walls and membranes leading to non-reversible uptake. The compounds may either be stored or metabolised at different rates. As shown by EPS and chlorophyll a proles, this uptake may be signicant at the sedimentwater interface and the layer below. In cases of strong hydrophobic binding and diffusion into intra-particle pores, the release of the compounds may be negligible, leading to observed irreversible sorption. This irreversible reaction may involve organic matter coated onto sediment particles, which in turn may enhance sorption for surface layers. In both instances the modelling may be extended by including irreversible sorption (as well as reversible) or bio-accumulation. Reversible sorption is controlled by KOM (as described above) while irreversible sorption and active uptake are characterised by rstorder sorption (organic matter-dependent), kIRR , and active uptake, kASS , rate constants respectively. For irreversible sorption: DSed 2 CMOC CMOC = t 1 + fOM KOM x2 (kIRR fOM CMOC )

(7)

The equation is similar for active uptake with the exception of the organic matter-dependence of the uptake rate constant. The rst-order active uptake rate may be described as: dCMOC = kASS CMOC (x, t) dt (8)

and the total mass of MOC bio-accumulated or biosorbed at the end of the experiment is calculated as the volume of pore water (Vp ) times the difference in pore water concentrations:
te 2 mm

n=

CMOC VP = VP kASS
0 0 mm

CMOC (x, t)dxdt

(9) with the total length of the experiment te (s) and VP the volume of pore water (litre), the mass of MOC assimilated, n (g) during the total length of the experiment, and the total change in pore water concentration with time, CMOC . This is applied to the top 2 mm below the sedimentwater interface for active uptake while it is applied to the whole sediment column for organic matter-dependent irreversible
Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

sorption. A SIMPLEX optimisation36 of KOM and rate constants is performed as described previously. As an example, results for lindane in Channel 1 for the experiment in light conditions were modelled following these assumptions. The resulting proles (Fig 6d) showed an excellent agreement of pore water concentrations (RMS = 0.72) for active uptake, with optimised values of 1100 litre kg1 for KOM and 1.04 106 s1 for kASS . However, it slightly overpredicted the whole sediment concentration (RMS = 0.60) in the top 2-mm layer, which could be explained by degradation in the biolm. The optimised active uptake was lower than the mean value obtained by Headley et al 1 of 3.83 105 s1 for a microbial biolm. An optimised KOM value of 4006 litre kg1 and irreversible sorption rate constant of 6.07 104 s1 provided a better agreement for whole sediment concentrations in top sections (RMS = 0.53) but did not allow enough diffusion to deeper layers. The predicted pore water prole underestimated the observed concentrations (RMS = 0.65). Deviations between observed and predicted proles in these conditions were generally similar to those obtained after using two degradation rate constants. It is not possible here to conclude which processes are occurring, but the limitations in using a simple reversible sorption process to describe the interactions with the sediment and biolm are identied. Simazine degradation rates in solution in the absence of sediment were higher than those found from the optimisations using the model with a single degradation rate for the sediment prole (Table 4). However, in Channel 1, an optimisation of two distinct degradation rates applied to the oxic and anoxic compartments produced a half-life of approximately 5 days for the oxic and 29 days for the anoxic section. A higher degradation rate for the 2 mm-deep oxic layer is consistent with the constants obtained for the overlying water, as similar conditions and micro-organism communities are expected in both compartments. Experiments under different redox status33 showed a similar trend, with an accelerated degradation of simazine under aerobic conditions (half-life of 15.1 days) and slower transformation under anaerobic conditions (half-life of 36.5 days). The optimised value for the single kdeg in the present work is close to the value obtained under oxic conditions. In the case of two distinct degradation rates, the value found for anaerobic conditions is close to the one obtained in this work for anoxic conditions. Simazine degradation occurs mainly in the overlying water. The activity of the bacterial and algal biolm under different conditions does not have a major impact on simazine degradation. It is likely that degradation is biotic, since the concentration remains constant for the rst 7 days. This also corresponds to the period of development of the algal and bacterial biolm at the sedimentwater interface. In the case of degradation by microbial communities, the growth of
Pest Manag Sci 60:417433 (online: 2004)

the biolm providing oxygen and algal exudates may enhance microbial or more generally heterotrophic activity at the sedimentwater interface42 and possibly degradation. EPS and chlorophyll a proles from the experiment in dark conditions (Fig 2) exhibited increases towards the sedimentwater interface, demonstrating the presence of a biolm in dark conditions. In addition, a dark-grown biolm may see its heterotrophic activity increased by autotrophs turning to heterotrophy for survival. The faster degradation of lindane in the oxic layer (Table 4) and much slower in the anoxic layer identied by modelling is contrary to expectations, as other workers found preferential degradation under anaerobic conditions for lindane.35 However, only low concentrations of lindane penetrated deeper in the sediment and may not have triggered biodegradation in the present experiment. These oxic-layer-specic half-lives are in agreement with previous results with river water (Table 6) and with the rapid losses from overlying waters seen in the dark. The presence at the sedimentwater interface or oxic layer of heterotrophic micro-organisms such as lamentous cyanobacteria43 may be responsible for this accelerated degradation.

7 CONCLUSIONS Diffusive transport and degradation of the compounds in sediments were complex and not described using a single distribution coefcient. The modelling procedure gave some insight into the factors that affect the concentrations in pore waters and sediments. The presence of an algal biolm is shown to concentrate the compounds at the surface, particularly for lindane. For experiments in the dark and light, the high concentrations in the surface layers do not respond quickly to reductions in the concentrations in the associated pore water or overlying water. Concentration measurements on river sediments, which are often made on mixed sediment samples to a depth of 5 cm, were shown to give a poor indication of the concentration in the surface layer and to depend on the history of exposure of the sediment to the pesticides. The ume experiments were successful in the analysis of the diffusive transport of simazine and lindane across the sedimentwater interface and within a 50-mm-deep sediment bed, allowing the novel measurement of ne-resolution whole sediment and pore water concentration-depth proles after a 42day period. The experiments showed the diffusion of lindane and simazine to 3050 mm in the sediment. Pore water concentrations were generally higher than in bulk solution at the end of the experiments, with a decreasing concentration gradient towards the sedimentwater interface for both compounds. This gradient may be attributed to back-diffusion into the overlying water, enhanced sorption in top sections, or specically faster degradation in oxic compared with anoxic compartments. Experimentally
431

IJ Allan et al

determined distribution coefcients for each sediment section can be modelled as a function of depth using power-law relationships, supported by organic matter, particle size, specic surface area vertical distributions and enhanced sorption at the SWI due to the biolm growth. For simazine, optimum agreement with the model was obtained in light conditions without invoking degradation in the sediment. However, the agreement was improved for the experiment in the dark if degradation in the sediment was permitted, although there was no improvement when differential degradation in the oxic and anoxic compartments was introduced. In contrast, the best agreement with the model for lindane in all experiments was obtained with differential degradation in the sediment.

ACKNOWLEDGEMENTS The authors wish to thank the Natural Environment Research Council, UK (NERC) for the CASE studentship to IJ Allan and for nancial support of this work.

REFERENCES
1 Headley JV, Gandrass J, Kuballa J, Peru KM and Gong Y, Rates of sorption and partitioning of contaminants in river biolm. Environ Sci Technol 32:39683973 (1998). 2 Spurlock FC and Biggar JW, Effect of naturally occurring soluble organic matter on the adsorption and movement of simazine [2-chloro-4,6-bis(ethylamino)-s-triazine] in Hanford sandy loam. Environ Sci Technol 24:736741 (1990). 3 Meakins NC, Bubb JM and Lester JN, The mobility, partitioning and degradation of atrazine and simazine in the salt marsh environment. Mar Pollut Bull 30:812819 (1995). 4 Lawrence JR, Kopf G, Headley JV and Neu TR, Sorption and metabolism of selected herbicides in river biolm communities. Canad J Microbiol 47:634641 (2001). 5 Karickhoff SW, Organic pollutant sorption in aquatic systems. J Hydraulic Eng 110:707735 (1984). 6 Pignatello JJ and Xing B, Mechanisms of slow sorption of organic chemicals to natural particles. Environ Sci Technol 30:111 (1996). 7 Koelmans AA, Hubert E, Koopman HW, Portielje R and Crum SJH, Modeling the vertical distribution of carbendazim in sediments. Environ Toxicol Chem 19:793800 (2000). 8 Mersie W, McNamee C, Seybold CA and Tierney DP, Diffusion and degradation of atrazine in a water/sediment system. Environ Tox Chem 19:20082014 (2000). 9 Elder JF, James RV and Steuer JJ, Mobility of 2,2 ,5,5 tetrachlorobiphenyl in model systems containing bottom sediments and water from the Lower Fox river, Wisconsin, USA. J Great Lakes Res 22:697706 (1996). 10 House WA, Denison FH, Warwick MS and Zhmud BV, Dissolution of silica and the development of concentration proles in freshwater sediments. Applied Geochem 15:425438 (2000). 11 Woodruff SL, House WA, Callow ME and Leadbeater BSC, The effects of a developing biolm on chemical changes across the sedimentwater interface in a freshwater environment. Int Rev Hydrobiol 84:509532 (1999). 12 Daniels WM, House WA, Rae JE and Parker A, The distribution of micro-organic contaminants in river bed-sediment cores. Sci Tot Environ 253:8192 (2000). 13 Daniels WM, House WA, Zhmud BV, Rae JE and Parker A, Diffusive movement of simazine and lindane in river-bed sediment. Pestic Sci 54:212222 (1998). 432

14 House WA, Long JLA, Rae JE, Parker A and Orr DR, Occurrrence and mobility of the insecticide permethrin in rivers in the Southern Humber catchment, UK. Pest Manag Sci 56:597606 (2000). 15 Woodruff SL, House WA, Callow ME and Leadbeater BSC, The effects of biolm on chemical processes in surcial sediments. Freshwater Biol 41:7389 (1999). 16 House WA, Denison FH, Smith JT and Armitage PD, An investigation of the effects of water velocity on inorganic phosphorus inux to a sediment. Environ Poll 89:263271 (1995). 17 Dubois M, Gilles KA, Hamilton JK, Rebers PA and Smith F, Colorimetric method for determination of sugars and related substances. Anal Chem 28:350356 (1956). 18 Underwood GJC, Paterson DM and Parkes RJ, The measurement of microbial carbohydrate exopolymer from intertidal sediments. Limnol Oceanogr 40:12431253 (1995). 19 Bascomb CL, Rapid method for the determination of cationexchange capacity of calcareous soils. J Sci Food Agric 15:821823 (1964). 20 Brunauer S, Emmett PH and Teller E, Adsorption of gases in multimolecular layers. J Amer Chem Soc 60:309319 (1938). 21 Crank J, The mathematics of diffusion, 2nd edn, Clarendon Press, Oxford (1975). 22 Boudreau BP, Diagenetic models and their implementation modelling transport and reactions in aquatic sediments, SpringerVerlag, Berlin (1997). 23 Zhmud BV, House WA and Denison FH, Release kinetics and concentration prole of dissolved silicon in compacted sediment. J Chem Soc, Faraday Trans 93:34733478 (1997). 24 Akgerman A and Gainer JL, Predicting gas-liquid diffusivities. J Chem Eng Data 17:372377 (1972). 25 van der Wielen LAM, Zomerdijk M, Houwers J and Luyben KCA, Diffusivities of organic electrolytes in water. Chem Eng J 66:111121 (1997). 26 Capel PD and Eisenreich SJ, Sorption of organochlorines by lake sediment porewater colloids, in Aquatic humic substances, ed by Suffet IH and McCarthy P, American Chemical Society, Washington, DC, pp 185208 (1989). 27 Niesner R and Heintz A, Diffusion coefcients for aromatics in aqueous solution. J Chem Eng Data 45:11211124 (2000). 28 Sastri SRS, Mohanti S and Rao KK, Liquid volume at normal boiling point. Canad J Chem Eng 74:170172 (1996). 29 Reid RC, Prausnitz JM and Poling BE, The properties of gases and liquid, McGraw Hill International, New York (1988). 30 Mackay D, Shiu W and Ma K, Illustrated handbook of physicochemical properties and environmental fate of organic chemicals, Vol 5, Lewis Publishers, New York (1997). 31 Skempton AW, Notes on the compressibility of clays. Quart J Geol Soc London 100:119135 (1943). 32 McVay M, Townsend F and Bloomquist D, Quiescent consolidation of phosphatic waste clays. J Geotech Eng 112:10331049 (1986). 33 Vink JPM and van der Zee SEATM, Effect of oxygen status on pesticide transformation and sorption in undisturbed soil or lake sediment. Environ Tox Chem 16:609616 (1997). 34 Wolfe NL and Macalady DL, New perspectives in aquatic redox chemistry: abiotic transormations of pollutants in groundwater and sediments. J Contam Hydrol 9:1734 (1992). 35 Middeldorp P, Jasper M, Zehnder A and Shraa G, Biotransformation of alpha-, beta-, gamma- and delta-hexachlorohexane under methanogenic conditions. Environ Sci Technol 30:23452349 (1996). 36 Miller JN and Miller JC, Statisitics and chemometrics for analytical chemistry, 4th edn, Pearson Education Limited, Harlow, UK (2000). 37 Reible DD, Popov V, Valsaraj KT, Thibodeaux LJ, Lin F, Dickshit M, Todaro MA and Fleeger JW, Contaminant uxes from sediment due to tubicid oligochaete bioturbation. Water Res 30:704714 (1996). 38 Comber SDW, Abiotic persistence of atrazine and simazine in water. Pestic Sci 55:696702 (1999).

Pest Manag Sci 60:417433 (online: 2004)

Transport and distribution of lindane and simazine in river sediment

39 Stangroom SJ, Collins CD and Lester JN, Abiotic behaviour of organic micropollutants in soils and the aquatic environment. A review: II. Transformations. Environ Technol 21:865882 (2000). 40 Chen G and Strevett K, Impact of bacterial extracellular polymers on lindane transport. Environ Eng Sci 18:191203 (2001). 41 Ju Y-H, Chen T-C and Liu JC, A study on the biosorption of lindane. Colloids Surf B: Biointerfaces 9:187196 (1997). 42 Romani AM and Sabater S, Effect of primary producers on the heterotrophic metabolism of a stream biolm. Freshwater Biol 41:729736 (1999). 43 Kuritz T and Wolk CP, Use of lamentous cyanobacteria for biodegradation of organic pollutants. Appl Environ Microbiol 61:234238 (1995).

Pest Manag Sci 60:417433 (online: 2004)

433