Environ. Sci. Technol. 2002, 36, 2205-2212
Effect of Exposure History on Microbial Herbicide Degradation in an Aerobic Aquifer Affected by a Point Source N I N A T U X E N , * ,† J U L I A R . D E L I P T H A Y , ‡,§ HANS-JØRGEN ALBRECHTSEN,† JENS AAMAND,‡ AND POUL L. BJERG† Environment & Resources DTU, Groundwater Research Centre, Technical University of Denmark, DK-2800 Lyngby, Denmark, and Department of Geochemistry, Geological Survey of Denmark and Greenland, DK-2400 Copenhagen NV, Denmark
The effects of in situ exposure to low concentrations (micrograms per liter) of herbicides on aerobic degradation of herbicides in aquifers were studied by laboratory batch experiments. Aquifer material and groundwater were collected from a point source with known exposure histories to the herbicides mecoprop (MCPP), dichlorprop, BAM, bentazone, isoproturon, and DNOC. Degradation of the phenoxy acids, mecoprop and dichlorprop, was observed in five of six sampling points from within the plume. Mecoprop was mineralized, and up to 70% was recovered as 14CO2. DNOC was degraded in only two of six sampling points from within the plume, and neither BAM, bentazone, nor isoproturon was degraded in any sampling point. A linear correlation (R2 g 0.83) between pre-exposure and amount of herbicide degraded within 50 days was observed for the phenoxy acids, mecoprop and dichlorprop. An improved model fit was obtained from using Monod degradation kinetics compared to zero- and first-order degradation kinetics. An exponential correlation (R2 g 0.85) was also found between numbers of specific phenoxy acid degrading bacteria and pre-exposure. Combination of these results strongly indicates that the low concentration exposure to phenoxy acids in the aquifer resulted in the presence of acclimated microbial communities, illustrated by the elevated numbers of specific degraders as well as the enhanced degradation capability. The findings support application of natural attenuation to remediate aerobic aquifers contaminated by phenoxy acids from point sources.
Introduction Herbicides have been widely used during the past 50 years, resulting in numerous findings of these substances in groundwater (1-5). This contamination may be a result of leaching from agricultural areas, but recent investigations show that point sources (landfills, machine pools, market * Address correspondence to this author at the Technical University of Denmark, Environment & Resources DTU, Bldg. 115, 2 Bygningstorvet, DK-2800 Kgs. Lyngby, Denmark. Phone: (+45) 45 25 15 95; fax: (+45) 45 93 28 50; e-mail:
[email protected]. † Technical University of Denmark. ‡ Geological Survey of Denmark and Greenland. § Present address: Department of General Microbiology, University of Copenhagen, DK-1307 Copenhagen K, Denmark. 10.1021/es0113549 CCC: $22.00 Published on Web 04/17/2002
2002 American Chemical Society
gardens, etc.) may also contribute (6-8). Typical herbicide concentrations in groundwater from landfills are 10-250 µg/L (8-10). These concentrations exceed the limit level set by the European Union of 0.1 µg/L by several orders of magnitude. Contrary to that observed from diffuse sources, the extent of water affected from point sources is limited, thus making it possible and relevant to assess remedial actions. One remediation strategy that has gained much interest during recent years is natural attenuation (11). The basic principle of natural attenuation is that naturally occurring processes such as biodegradation, dispersion, dilution, adsorption, and volatilization reduce the mass of critical pollutants to eliminate risk for down-gradient groundwater (12). A crucial point in the assessment of natural attenuation is whether the specific compounds in question are degraded. So far, natural attenuation has not been applied to herbicide contaminations, but it appears from the literature that phenoxy acids are some of the most readily degraded herbicides (13-18), indicating a potential for using natural attenuation in cases of phenoxy acid contamination. However, most of the reported results originate from laboratory investigations, and knowledge on phenoxy acid degradation in aquifers is sparse. To maintain sufficient degradation rates for natural attenuation, a microbial population capable of degrading the target compounds must develop. It is presently unknown whether the low micrograms per liter concentrations of herbicides, such as found in groundwater, can support the growth of specific herbicide degraders in situ. Alexander (19) reported that typical threshold levels which support growth are in the range of micrograms per liter for many contaminants. For the phenoxy acid 2,4-D, a threshold level of 1-2 µg/L has been determined by laboratory batch experiments (20, 21). However, threshold levels determined in batch experiments cannot directly be transferred to aquifer flow systems, where a flux of low concentrations passing the bacteria attached to the sediment over time will result in a relatively high accumulated exposure to the compounds. Enhanced degradation of herbicides, possibly due to an increase in the population of microbial degraders, has been shown for very high herbicide concentrations in topsoil (milligrams to grams per kilogram levels) [for review, see Roeth (22)]. However, these studies focus on the potential negative impact of enhanced degradation on weed control and crop production. Degradation rates and extents of mineralization of 2,4-D and MCPA (e.g., refs 23-27), glyphosate (23), simazine (23), and bentazone (28) were increased after pre-exposure with the herbicides, whereas pre-exposure to simazine (29) and isoproturon (29, 30) had no effect and pre-exposure to atrazine caused differing results; enhanced degradation was found in one investigation (24) and not in another (29). Growth of microorganisms due to pre-exposure has also been reported in aquifers contaminated with petroleum hydrocarbons in milligrams per liter concentrations (31-33). No studies exist on the potential positive effect of enhanced degradation of low herbicide concentrations in aquifers on the environmental fate and the potential implications of natural attenuation. Prior to this study a natural gradient field injection was conducted in an aerobic aquifer (34). The herbicides mecoprop, dichlorprop, bentazone, DNOC, isoproturon, and the metabolite BAM (Table 1) and bromide as tracer were continuously injected 2 m below the groundwater table for a period of 216 days. The concentration of each herbicide was ∼40 µg/L in the groundwater immediately down-gradient VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2205
TABLE 1. Herbicides under Investigation
hydraulic conductivity of the aquifer is 2 × 10-4 m/s, and the groundwater velocity is ∼0.5 m/day (34). The aquifer is aerobic, with an oxygen content of 2-10 mg/L, and the pH changes over depth with an increase from 4.3 in the upper part to 7 in the lower part of the aquifer (34). The temperature of the aquifer is ∼10 °C. Sampling points (A1, A2, B1, B2, C1, and C2) with different exposure histories were selected within the area of the herbicide plume. All sampling points were selected along the center flow line with increasing distance from the injection wells (A ) 0.5 m, B ) 1.5 m, and C ) 4.5 m) at two different depths (1 ) upper and 2 ) lower). Additionally, four sampling points (NX1-4) were selected outside the plume area with no history of exposure (Table 2). The position of the sampling points and the extension of the contaminant plume, illustrated by the bromide plume, are shown in Figure 1. At the time of sampling (40 days after termination of the injection), the bromide plume was no longer present within the monitoring network of the field site. The aquifer material was collected as up to 110 cm cores with a stainless steel piston sampler (L ) 5.6 cm) modified after the procedure of Starr and Ingleton (37). The cores were stored at 4 °C until used, 98%) was added in concentrations of 23 µg/L. Control experiments (for the B, C, and NX sampling points) were set up as biologically active microcosms but amended with 100 mg/L HgCl2 to inhibit microbial activity. All microcosms (a total of 20 microbial active and 6 sterile microcosms) were incubated in the dark at 10 °C, corresponding to the aquifer temperature. During the 180 days of incubation water samples were collected periodically. Prior to sampling (or once a week), the microcosms were shaken for 2 min followed by 6 h of horizontal resting. At each sampling 7 mL of water was collected from the supernatant with sterilized syringes and needles. Sterile filtrated atmospheric air was added as compensation. Three milliliters of each sample was stored in a -18 °C freezer for subsequent HPLC analysis, whereas the remaining 4 mL was transferred in a closed syringe to the outer vial of a double-vial system. The inner vial, used as a 14CO trap, contained 1 mL of 2.5 M KOH, and when 0.2 mL 2 of 37% HCl was added to the outer vial, the double-vial was closed immediately and gently shaken. Tests with [14C]HCO3showed that no 14CO2 was lost from the outer vial during this procedure and that all 14CO2 in the sample was trapped in the KOH solution within 48 h (data not shown). Analytical Procedures. The pH and the O2 content were measured in the field in a miniature flow cell with a WTW Microprocessor pH meter and a WTW Microprocessor
TABLE 2. Characteristics of the Aquifer Material, Groundwater Chemistry, and Herbicide Exposure at Each Sampling Point exposed sampling point: distance from injection well (m): depth (m below surface): aquifer material TOC (mg of C/g of dwa) d50 (mm) surface area (m2/g) groundwater chemistry (mg/L) ClNO3- -N SO42- -S Na+ K+ Ca2+ Mg2+ O2 pH (-) exposure (µg) MCPP dichlorprop 2,4-D DNOC isoproturon bentazone BAM a
nonexposed
A1 0.5 5.25
A2 0.5 5.50
B1 1.5 5.25
B2 1.5 5.75
C1 4.5 5.75
C2 4.5 6.25
NX1 -8 5.00
NX2 -8 5.25
NX3 5 5.50
NX4 5 6.00
0.08 0.31 1.82
0.10 0.42 1.4
0.10 0.37 1.48
0.14 0.58 2.72
0.14 0.31 1.03
0.07 0.50 1.06
0.08 0.57 1.37
0.11 0.50 1.76
0.08 0.42 1.12
0.10 0.43 0.99
39.9 10.4 11.4 12.4 5.6 27.6 3.5 7.8 5.05
39.3 13.3 13.5 14.5 5.2 30.9 4.4 8.4 5.39
47.5 13.8 10.8 12.9 4.7 30.9 4.1 7.9 4.99
42 14 17.5 15.3 5.0 34.2 6.2 8.7 5.13
37.2 15 17.9 14.6 5.1 34.9 4.5 7.1 5.04
40.6 13.1 18.9 13.4 5.4 38.8 5.1 7 5.55
51 7.5 10 20.3 4.1 23.4 2.0 8.4 5.12
42.9 9.7 9.1 17.2 3.7 24.1 2.9 7.9 5.49
44 10 9.7 15.4 6.5 28.8 3.9 8.4 5.4
45 11.5 9.5 15.9 4.6 25.4 3.5 8.2 5.19
31 34 0 57 49 50 40
37 41 0 72 63 64 52
11 13 0 25 27 31 26
25 30 0 57 57 57 49
20 22 0 50 52 63 55
20 20 0 42 51 56 44
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
Dry weight of sample before removal of inorganic carbon.
FIGURE 1. Locations of aquifer and groundwater sampling points at the field site and position of the bromide tracer plume from the field injection. Sediment and groundwater sampling for this study was done 40 days after the injection was terminated and bromide was no longer present within the monitoring network. The upper panel is a plan view with the bromide plume seen from above, and the lower panel is a longitudinal cross section of the aquifer with the bromide plume seen along a central flow line. Note that sampling points NX1-4 are located outside the plume. oximeter. Oxygen in the headspace of the microcosms was quantified using a Chrompack Micro GC CP-2002P gas chromatograph with a thermal conductivity detector (39). Cations were measured with a Perkin-Elmer 5000 atomic absorption spectrophotometer using standard conditions. Anions were measured on a Dionex ion chromatograph DX120 (16). The sediment-bound organic carbon was
determined in a LECO oven after removal of inorganic carbon by treatment with 6% H2SO3 (40). Specific surface area was determined by the BET method using a Coulter surface area analyzer (SA 3100) with >99.999% pure N2. All samples were outgassed at 80 °C for 34 h. The amount of 14CO2, [14C]mecoprop, and [14C]-2,4-D was quantified by using a Packard TriCarb 2000 liquid scintillation counter after the addition of HiSafe 3 (Wallac) scintillation cocktail. Herbicides were analyzed on a Hewlett-Packard series 1100 HPLC system at wavelengths of 220 nm (BAM and bentazone), 205 nm (mecoprop, dichlorprop, and isoproturon), and 370 nm (DNOC) (34). The detection limits were ∼1 µg/L. The numbers of specific degraders in each sediment sample were enumerated by use of a Most Probable Number technique (MPN) as reported in ref 43. Five replicates of each sediment dilution (10-1-10-5) added in a minimal medium were incubated for 3 months with each of the phenoxy acids as the only carbon source. Samples were scored positive if mineralization exceeded 25%. For further details see ref 43. Modeling of the 10 degradation curves for each of the three phenoxy acids was performed using the flexible code Aquasim (41), which allows for a range of degradation model descriptions. Three degradation kinetics were tested: zeroand first-order degradation as well as single-type Monod degradation including growth but without inhibition or microbial decay. The Monod equation was used in the form presented by Simkins and Alexander (42)
-dC µmaxC(C0 + X0 - C) ) dt Ks + C
(1)
where C is the herbicide concentration in the water, C0 is the initial herbicide concentration, µmax is the maximum specific growth rate, Ks is the half-saturation constant for growth, and X0 is the initial biomass expressed as the amount of herbicide required to produce a certain population density. For the development of eq 1 the following mass balance was used, expressing the growth of the biomass X (42):
C0 + X0 ) C + X VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
(2) 9
2207
TABLE 3. Estimated Values of the Parameters in the Monod Degradation (Equation 1) mecoprop
A1 A2 B1 B2c C1 C2 NX1 NX2 NX3 NX4
dichlorprop
X0a (cells/g)
model X0b (µg/L)
µmax/Ks (×10-3 L/day‚µg)
X0 a (cells/g)
350 2200 2 110 70 170
