OUP user menu

Centimetre-scale vertical variability of phenoxy acid herbicide mineralization potential in aquifer sediment relates to the abundance of tfdA genes

Meriç Batıoğlu-Pazarbaşı, Jacob Bælum, Anders R. Johnsen, Sebastian R. Sørensen, Hans-Jørgen Albrechtsen, Jens Aamand
DOI: http://dx.doi.org/10.1111/j.1574-6941.2012.01300.x 331-341 First published online: 1 May 2012

Abstract

Centimetre-scale vertical distribution of mineralization potential was determined for 2,4-dichlorophenoxyacetic acid (2,4-D), 4-chloro-2-methylphenoxyacetic acid (MCPA) and 2-(4-chloro-2-methylphenoxy)propanoic acid (MCPP) by 96-well microplate radiorespirometric analysis in aquifer sediment sampled just below the groundwater table. Mineralization of 2,4-D and MCPA was fastest in sediment samples taken close to the groundwater table, whereas only minor mineralization of MCPP was seen. Considerable variability was exhibited at increasing aquifer depth, more so with 2,4-D than with MCPA. This suggests that the abundance of MCPA degraders was greater than that of 2,4-D degraders, possibly due to the fact that the overlying agricultural soil had long been treated with MCPA. Mineralization of 2,4-D and MCPA was followed by increased abundance of tfdA class I and class III catabolic genes, which are known to be involved in the metabolism of phenoxy acid herbicides. tfdA class III gene copy number was approximately 100-fold greater in samples able to mineralize MCPA than in samples able to mineralize 2,4-D, suggesting that tfdA class III gene plays a greater role in the metabolism of MCPA than of 2,4-D. Degradation rate was found to correlate positively with tfdA gene copy number, as well as with the total organic carbon content of the sediment.

Keywords
  • spatial variability
  • radiorespirometric analysis
  • tfdA diversity
  • TaqMan probe-based quantitative real-time PCR

Introduction

Phenoxy acid herbicides, which are widely used in agriculture to control broadleaf weeds, are characterized by a low tendency to sorb to soil and a relatively high water solubility. As a consequence, they may leach from agricultural soil to underlying groundwater, thereby threatening the use of groundwater as a drinking water resource. In the European Union, the concentration limit level for individual pesticides in groundwater is 0.1 μg L−1 (European Parliament & Council, 2006).

Soil and groundwater systems are complex environments exhibiting a high degree of spatial variability in their physical, chemical and biological properties. Topsoil is known to exhibit spatial variability in bacterial abundance (Becker et al., 2006; Philippot et al., 2009), herbicide mineralization potential (Gonod et al., 2003, 2006; Rodriguez-Cruz et al., 2009; Sjøholm et al., 2010), community structure (Franklin & Mills, 2003; Griffiths et al., 2003; Brad et al., 2008) and physico-chemical properties (Franklin & Mills, 2009; Baker et al., 2009). The potential of soil aggregates to mineralize 2,4-dichlorophenoxyacetic acid (2,4-D) can be extremely heterogeneous (Gonod et al., 2003) at millimetre scale, a phenomenon attributed to uneven distribution of degrading populations rather than variability in soil characteristics at this scale. Gonod et al. (2006) reported that inter-sample differences in 2,4-D mineralization increased when the spatial scale was reduced from field scale to microhabitat scale, concluding that 2,4-D mineralization in topsoil was only spatially structured at microhabitat scale.

Much less is known about the vertical variability of herbicide degradation potential in aquifers (Larsen et al., 2000; Albrechtsen et al., 2001; Wood et al., 2002); however, small-scale vertical variability of sorption and herbicide degradation has recently been reported in sediment samples from an unsaturated limestone aquifer and a sandy aquifer (Janniche et al., 2011), but apart from this study, to our knowledge, none have addressed small-scale spatial variability of phenoxy acid herbicide mineralization potential in subsurface environments.

The rapid transport of water and solutes in soil described as preferential flow originates either from the macropore flow or nonhomogeneous infiltration and wetting front instabilities (Bundt et al., 2001). The solutes bypass a large part of the soil matrix, and thus, pesticides or nutrients may become more mobile than anticipated and cause groundwater contamination (Flury et al., 1994; Stamm et al., 1998). However, microorganisms may also rapidly be transported along preferential flow paths within the aquifer (Goldscheider et al., 2006). The groundwater table and capillary fringe comprise the most important transition zone separating the unsaturated and saturated zones of an aquifer. Typically, this zone is characterized by high biodiversity and microbial activity (Goldscheider et al., 2006). Berkowitz et al. (2004) claimed that the combination of the capillary fringe and the region immediately below the water table is a unique interface in terms of natural geochemical and microbial conditions, however, and suggested that the concept of the capillary fringe should be replaced by the concept of the partially saturated fringe involving multiphase transport immediately above and below the water table. Other relevant transition zones are the interfaces between lithological units and hyporheic zones involving surface water and groundwater. The dynamics of invertebrate and higher fauna has been examined in these transition zones (Vervier & Gibert, 1991; Williams, 1993), and a few studies have focused on microbial communities in only heavy metal contaminated hyporheic zones (Feris et al., 2003, 2004). However, none have examined pesticide degraders below the groundwater table.

Phenoxy acid herbicides are initially degraded by oxygenases encoded by cadAB (Kitagawa et al., 2002) or tfdA-like genes (Fukumori & Hausinger, 1993a, b; Itoh et al., 2002). Bacteria harbouring tfdA-like genes are divided into three groups based on their phylogeny and catabolic gene diversity (Kamagata et al., 1997). The first group consists of copiotrophic bacteria belonging to β- and γ-subdivisions of proteobacteria harbouring the tfdA gene. The second group consists of oligotrophic, slowly growing strains belonging to the BradyrhizobiumAgromonasNitrobacterAfipia cluster of α-proteobacteria harbouring tfdAα and cad gene clusters (Kamagata et al., 1997; Itoh et al., 2002, 2004; Kitagawa et al., 2002). The third group consists of copiotrophic, rapidly growing Sphingomonas strains belonging to α-proteobacteria harbouring tfdA-like genes and the cadA gene (Itoh et al., 2004).

The first group is subdivided into three classes based on differences in their tfdA sequences (McGowan et al., 1998). TaqMan probe-based quantitative real-time PCR assay has been developed to detect and discriminate the three classes of tfdA genes in soils (Bælum & Jacobsen, 2009). Degradation of phenoxy acids such as 2,4-D, 4-chloro-2-methylphenoxyacetic acid (MCPA) (Bælum et al., 2008) and mecoprop (Rodriguez-Cruz et al., 2009) has been shown to be associated with the number of tfdA genes and/or transcripts present in the soil samples. Heterogeneous distribution of mecoprop degradation potential has recently been reported to correlate positively with the proliferation of these catabolic genes in an agricultural field (Rodriguez-Cruz et al., 2009). It is unclear whether herbicide degradation in subsurface environments is determined by the abundance of tfdA genes, however, and whether it can be linked to a depth-related variability in herbicide mineralization potential.

The purpose of this study was to determine the centimetre-scale vertical variability in phenoxy acid herbicide mineralization potential in sediment sampled just below the groundwater table and correlate this with tfdA gene copy number and total organic carbon (TOC) content. Knowledge of small-scale distribution of degradation activity is necessary to be able to accurately predict the fate of pesticides in groundwater aquifers.

Materials and methods

Sampling

Sediment was sampled from just below the groundwater table in an aquifer located near Fladerne Creek, Jutland, Denmark in June 2009. The site is a small tributary catchment area in an extended moorland plain with meltwater sediments dominated by coarse sand. The water flow in the aquifer is mainly horizontal, and the area has been used as farmland since 1920. MCPA is used on the overlying farmland and has been detected in the underlying groundwater (Jensen et al., 2004). The sediment was taken as a core from the groundwater table 2.5 m below ground surface (b.g.s.) to a depth of 3.25 m b.g.s. using a 6-cm diameter stainless steel piston sampler modified from Starr & Ingleton (1992). The core was kept at 4 °C in the dark until needed. For the study, the sediment core was divided into 25 3-cm slices with the top and bottom slices being discarded to prevent contamination. A circular template with sampling holes was positioned on the surface of each slice and 0.2-g sediment samples collected from three adjacent holes using a 1-mL tip-cut syringe for each compound (Fig. ). The sediment samples were directly transferred to 96-well microplates (Sarstedt, Nümbrecht, Germany) for quantification of mineralization potential. Three 0.2 g sediment samples from each depth were stored at −20 °C for molecular analysis, and 5 g of sediment from each depth was dried at room temperature for analysis of pH and TOC.

Overview of the sampling approach. A circular template with 21 holes was placed on the surface of each sediment slice and 0.2 g sediments then sampled from three adjacent holes for each compound and DNA extraction using a 1-mL tip-cut syringe. C, control; BA, benzoic acid; I-3-AA, indole-3-acetic acid; 2,4-D, 2,4-dichlorophenoxyacetic acid; MCPA: 4-chloro-2-methylphenoxyacetic acid; MCPP: (R)-2-(4-chloro-2-methylphenoxy)propanoic acid.

Chemicals

The analyses were performed using radioactively labelled (ring-U-[14C]-benzoic acid (> 99% purity; Sigma-Aldrich, St. Louis, MO), indole-3-acetic acid-carboxy-14C (> 99% purity; Sigma-Aldrich), ring-U-[14C]-2,4-D (≥ 98% purity; Sigma-Aldrich), ring-U-[14C]-MCPA (> 95% purity; Izotop, Budapest, Hungary) and ring-U-[14C]-2-(4-chloro-2-methylphenoxy)propanoic acid) (> 95% purity; Izotop). The molecular structure of the compounds studied is shown in Fig. .

Molecular structure of the compounds investigated.

96-well microplate radiorespirometric analysis

The 0.2 g sediment samples were transferred to 96-well microplates, and 20 μL of a stock solution containing the 14C-labelled compound (125 000 DPM mL−1) and 10 mg L−1 of the unlabelled compound then added at a final concentration of 1 mg kg−1 sediment. To trap the 14CO2 produced, sealing tape containing Ca(OH)2-soaked cellulose filters was adhered to the top of the microplates. The plates were incubated at 15 °C and the sealing tape replaced at defined intervals. With 2,4-D, MCPA and 2-(4-chloro-2-methylphenoxy)propanoic acid (MCPP), the sealing tape was changed once weekly for 108 days. With benzoic acid (BA) and indole-3-acetic acid (I-3-AA), the sealing tape was replaced more frequently. The 14CO2 trapped on the cellulose filters was quantified by autoradiography followed by digital image analysis as described by Johnsen et al. (2009). Abiotic controls containing 20 μL sodium azide (69 mg mL−1) were included to monitor abiotic loss of radioactivity from the wells.

DNA extraction

DNA was extracted from three adjacent 0.2 g sediment samples using the MoBio Ultraclean soil DNA extraction kit (Cat no 12800-50, Carlsbad, CA) as described by the manufacturer, although with an additional bead beating step [three 30-s pulses at speed 4 on the FastPrep FP 120 instrument (Qbiogene, Inc., CA)]. DNA was also extracted after 2,4-D and MCPA mineralization following the same procedure.

SYBR Green-based quantitative real-time PCR

All PCR was performed on a Bio-Rad iCycler iQ5 real-time PCR instrument (Bio-Rad Laboratories, Inc.).

For 16S rRNA gene quantification, the thermo-cycling programme was started with 95 °C for 10 min, followed by 50 cycles of 95 °C for 45 s, 55 °C for 30 s and 72 °C for 1 min, and a final step at 72 °C for 6 min. Melting curve profiles of the PCR products were constructed using 71 cycles of 60–95 °C for 30 s increasing by 0.5 °C at every cycle, and 72 °C for 45 s. Tenfold serial dilutions of plasmid DNA with a 16S rRNA gene insert (GenBank accession no. JF523523) were used as standards for the quantitative real-time PCR.

For tfdA quantification, the thermo-cycling programme was started with 94 °C for 10 min, followed by 50 cycles of 94 °C for 30 s, 62 °C for 30 s, 72 °C for 1 min and a final step at 72 °C for 7 min. Melting curve profiles were determined for the PCR products using 77 cycles from 64 to 94.4 °C for 30 s increasing by 0.4 °C at every cycle, and 72 °C for 45 s. The melting curves were used to verify the presence of the specific real-time PCR product. Plasmid DNA including tfdA class I gene fragment was serially diluted for use as standards for the SYBR Green-based quantitative PCR, as described previously by Bælum & Jacobsen (2009).

Amplification was carried out using Maxima™ SYBR Green qPCR master mix (Fermentas International Inc, ON, Canada) containing 0.4 μM of each primer (341F-518R and tfdA-215 bp primer sets, Table ) and 1 μg μL−1 bovine serum albumin (New England Biolabs, Ipswich, MA). One microlitre of DNA extract (10–50 ng) was added to 19 μL of the PCR master mix in each well of the 96-well reaction plates. Quantification was carried out in triplicate for each sample.

View this table:

Primers and probes used for PCR

Oligo nameTarget geneOligo (5′–3′)Fragment size (bp)Annealing temp. (°C)
341F-518R16S rRNA geneF: CCT ACG GGA GGC AGC AG20055
R: ATT ACC GCG GCT GCT GG
tfdA – 81 bptfdA c I–IIIF: GAG CAC TAC GC(A/G) CTG AA(T/C) TCC CG8162
R: (G/C)AC CGG (A/C)GG CAT (G/C)GC ATT
tfdA – 215 bptfdA c I–IIIF: GAG CAC TAC GC(A/G) CTG AA(T/C) TCC CG21564
R: GTC GCG TGC TCG AGA AG
tfdA c ItfdA c IFAM-TTG CGC TTC CGA ATA GTC GGT GTC-BBQ62
tfdA c IItfdA c IICy5-CGT TGA CTT TCA GAA TAC TCT GTG TCG CCA-BBQ62
tfdA c IIItfdA c IIIYAK-TTG ACT TTC AGA ATA GTC CGT ATC GCC AAG-BBQ62
  • F, forward primer; R, reverse primer; c, class.

TaqMan probe-based quantitative real-time PCR

Quantification of tfdA classes I, II and III was performed by TaqMan® probe-based quantitative real-time PCR on a Bio-Rad iCycler iQ5 real-time PCR instrument. The thermo-cycling programme comprised 95 °C for 5 min, followed by 50 cycles of 95 °C for 30 s, 62 °C for 90 s and a final step at 72 °C for 7 min.

Standard series were constructed by cloning the amplified tfdA genes extracted from DNA of Cupriavidus necator JMP134 (class I), Burkholderia sp. RASC (class II) and Cupriavidus sp. TFD40 (class III) into PCR 2.1-TOPO cloning vectors (Invitrogen, Carlsbad, CA) following the manufacturer's instructions. Recombinant plasmid DNA was amplified in chemically competent Escherichia coli cells and extracted using the UltraClean™ 6 Minute Mini Plasmid Prep Kit (MoBio Laboratories). Plasmid DNA including tfdA class I, class II and class III gene fragments was mixed, serially diluted and used as standards for TaqMan® probe-based quantitative real-time PCR as described previously (Bælum & Jacobsen, 2009).

Amplification was carried out using iQ™ Supermix (Bio-Rad Laboratories) containing 0.4 μM of each primer (tfdA-81bp primer set; Table ), 0.1 μM of each probe (tfdA class I, II and III, Table ) and 1 μg μL−1 bovine serum albumin (New England Biolabs). One microlitre of DNA extract (10–50 ng) was added to 19 μL of the PCR master mix in each well of the 96-well reaction plates. Quantification was performed in triplicate for each sample.

pH and TOC analysis

Sediment pH was measured on 2 g of dried sediment mixed with 10 mL of 0.01 M CaCl2 (Metrohm 716 DMS Titrino, Herisau, Switzerland) (International Organization for Standardization, 1994). TOC was determined using 1 g of dried sediment that had been ground and sieved (250 μm mesh). CaCO3 was removed using HCl and TOC then determined by combustion in a Leco CS-200 oven (St. Joseph).

Data analysis

The correlation between time to achieve 10% mineralization and (I) tfdA gene copy number and (II) TOC was determined by Spearman rank correlation analysis.

Results

Centimetre-scale vertical variability in mineralization potential

Mineralization of BA – used here as a general measure of bacterial activity – was rapid and followed a similar pattern in all samples (Fig. a). Ten percent mineralization was achieved in < 1 day at all depths, with the time needed being only slightly longer at the deepest depths than at the shallowest depths (Fig. a). Mineralization of I-3-AA – a plant hormone with functional similarities to the phenoxy acid herbicides – was also rapid in the uppermost sediments, but slower at the deeper depths (Figs b and b), the time needed to achieve 10% mineralization being approximately 1 day in the upper sediment increasing to 1.5–3 days at a depth of 3.19 m b.g.s.

Mineralization of (a) benzoic acid, (b) I-3-AA, (c) 2,4-D and (d) MCPA in the three uppermost and lowermost sediment core slices. (♦) 2.53, (▲) 2.56, (◼) 2.59, (◊) 3.13, (△) 3.16, and (◻) 3.19 m b.g.s.

The time needed to achieve 10% mineralization vs. depth in the aquifer sediment core for (a) benzoic acid, (b) I-3-AA, (c) 2,4-D and (d) MCPA. > indicates that the duration exceeded 108 days.

Mineralization of 2,4-D was much slower than that of BA and I-3-AA, especially in the samples from the deeper parts of the core (Fig. c). Ten percent mineralization was achieved in < 40 days at all sample depths from 2.53 to 2.68 m b.g.s., while only a few of the samples from the depth range 2.68–2.99 m b.g.s. were able to mineralize 2,4-D, and at depths below 2.86 m b.g.s., it took longer than 108 days to achieve 10% mineralization, an exception being the sample from 3.13 m b.g.s. (Fig. c).

With MCPA, mineralization was fastest in the upper parts of the sediment (2.53–2.62 m b.g.s.), where all samples were able to mineralize the compound (Fig. d). At greater depths, MCPA only mineralized in some samples, and the time needed to achieve 10% mineralization was generally longer (Figs d and d). MCPA mineralized in more samples than did 2,4-D, and the mineralization distribution pattern differed from that of 2,4-D (Fig. c and d). None of the samples achieved 10% mineralization of MCPP within the 108-day incubation period (data not shown).

Quantitative PCR of 16S rRNA gene and total tfdA genes

The total number of bacteria expressed in terms of 16S rRNA gene copy number was approximately the same at all depths, ranging from 7.29 × 105 to 5.39 × 106 g−1 sediment (Fig. a). Copy number had increased at all depths after 2,4-D exposure during 108 days. In sediment samples able to mineralize 2,4-D, 16S rRNA gene copy number increased to the range 3.75 × 107–8.75 × 107 g−1 but was slightly lower in the samples in which 2-4-D did not mineralize (Fig. a). A similar pattern was observed for the sediments exposed to MCPA, where 16S rRNA gene copy number was in the range 1.91 × 107–1.81 × 108 g−1 in the samples able to mineralize MCPA as compared with 3.07 × 106–4.56 × 107 g−1 in the samples unable to mineralize MCPA (Fig. a).

Vertical distribution of (a)16S rRNA and (b) total tfdA gene number in the aquifer sediment core before and after mineralization of 2,4-D and MCPA. < indicates copy number < 100. The error bars in (a) represent the standard errors of the mean for sediment triplicates before mineralization. For active and nonactive samples after mineralization, 16S rRNA gene numbers were shown for only one location at each depth along the core to prevent the confusion. tfdA gene numbers were shown in triplicates for only active samples after mineralization.

Total tfdA gene copy number was < 102 g−1, i.e. below the detection limit for nonincubated sediments. This was also the case for the samples in which 2,4-D and MCPA did not mineralize during the 108-day experiment (Fig. b). In contrast, total tfdA copy number was in the range 1.56 × 103–5.08 × 104 g−1 following exposure to 2,4-D and was even higher following exposure to MCPA – 1.50 × 104–1.08 × 107g−1 (Fig. b).

Quantitative PCR of tfdA gene classes

As with total tfdA genes, specific class I and class III genes only appeared in the samples that were able to mineralize the two herbicides (Fig. a and b). In the samples that mineralized 2,4-D, gene copy number was 3.05 × 102–6.50 × 103 g−1 for class I and 1.46 × 103–1.54 × 104 g−1 for class III. In samples that mineralized MCPA, gene copy number was an average of 10-fold greater for class I (5.68 × 103–1.27 × 105 g−1) and 100-fold greater for class III (5.79 × 103–1.36 × 106 g−1). Because of a problem during the DNA extraction procedure, copy number could not be determined for 16S rRNA gene, total tfdA and tfdA classes I and III for 2,4-D at 3.13 m b.g.s.

Vertical distribution of (a) tfdA class I genes and (b) tfdA class III genes (c) the ratio of tfdA class III to class I genes in the aquifer sediment core after mineralization of 2,4-D and MCPA. < indicates copy number < 100.

The depth distribution of the tfdA class III:I ratio through the core differed between the two herbicides. The class III:I ratio was similar in the upper part of the core (3 : 1) for both herbicides but increased with depth in the samples able to mineralize MCPA (20 : 1) (Fig. c). Class II genes were also detected at several depths in the upper part of the core, but only after mineralization of MCPA (data not shown).

Physico-chemical properties

TOC ranged from 0 to 0.5 mg g−1 (Fig. ). In general, TOC was relatively high in the upper parts of the core and decreased with depth, although a slight increase was observed between 3.13 and 3.19 m b.g.s. pH ranged from 5.1 to 5.8.

Vertical distribution of TOC and pH in the aquifer sediment core.

Data analysis

The Spearman rank correlation coefficients for the time needed to achieve 10% mineralization vs. total tfdA, tfdA class I and III were negative as seen in Table (P < 0.001). The time to 10% mineralization was also correlated with sediment TOC with negative correlation coefficients as indicated in Table (P < 0.05). Thus, degradation rate determined based on time to 10% mineralization was correlated positively with both tfdA gene and sediment TOC along the core at significance levels of 0.1% and 5%, respectively.

View this table:

Spearman rank correlations between time to 10% mineralization and (I) the copy number of tfdA gene (P < 0.001) and (II) TOC (P < 0.05)

After 2,4-D mineralizationAfter MCPA mineralizationOutcome
Spearman correlation coefficient (Rho)
123123
I. Time to 10% mineralization vs. tfdA gene copy numberTotal tfdA−0.95−0.93−0.92−0.92−0.83−0.86Negative correlation
tfdA class I−0.97−0.92−0.93−0.97−0.87−0.88
tfdA class III−0.96−0.94−0.92−0.93−0.84−0.87
II. Time to 10% mineralization vs. TOC−0.6−0.5−0.6−0.6−0.6−0.6Negative correlation
  • The spearman rank correlation analyses were performed for adjacent subsamples (1,2,3) from each depth.

Discussion

2,4-D and MCPA mineralization potential in the saturated zone close to the groundwater table within a Danish sandy aquifer exhibited depth-related variability on a centimetre-scale. BA and I-3-AA mineralized at all depths, thus precluding the occurrence of bacterially inactive zones along the sediment core. In contrast to the vertical variability in mineralization potential following exposure to 2,4-D and MCPA, 10% mineralization of MCPP was not achieved in any of the samples during the 108-day experiment. This could be attributable to the fact that as MCPP has not been applied to the farmland overlying the aquifer in the past, the degrader population was not adapted to mineralization of MCPP. With 2,4-D and MCPA, mineralization occurred most rapidly in the sediment closest to the groundwater table. At greater depths, the time to achieve 10% mineralization increased for both compounds, and some samples were unable to mineralize the compounds. Our study thus demonstrates rapid and evenly distributed mineralization immediately below the groundwater table, but increasing variability with increasing sediment depth. This heterogeneous distribution of actively mineralizing sediment could be attributable to differences in exposure of the microorganisms above and below the groundwater table to herbicides, nutrients and oxygen in water infiltrating from the overlying farmland via preferential flow. In fact preferential flow is known to occur in the vadose zone above this sandy aquifer, where infiltration water passes through vertical canalized fingers after intensive precipitation (Jacobsen et al., 1998). The infiltrating water carrying phenoxy acid herbicides and microorganisms moves rapidly through the vertical canalized fingers along the vadose zone and then reaches the groundwater table. However, along the saturated zone, infiltrating water can move slowly and irregularly through the pores and mixes with the resident water. Thus, the indigenous population exposed to heterogeneously distributed phenoxy acids might have tendency to create hotspots where the contamination level is higher. In our study, the rapid mineralization in the region immediately below the groundwater table and increasing variability with depth might be attributed to the biological filter function of this transition zone. This is in agreement with a previous study addressing that the complex combination of water flow patterns in the sediment located close to the water table may cause highly geochemically and microbiologically active zones, not only in the vadose zone but also in the upper portion of the groundwater zone (Berkowitz et al., 2004).

The vertical distribution in mineralization potential correlated with that of TOC. This is in line with previous reports indicating that the spatial distribution of easily degradable organic carbon may partly control the spatial distribution of 2,4-D mineralization potential (Gonod et al., 2003) and that TOC and microbiological parameters such as substrate-induced respiration and fluorescein diacetate hydrolysis are important determinants of variability in MCPA dissipation rate (Juhler et al., 2008). Similarly, spatial heterogeneity in the rate of isoproturon degradation has been shown to correlate with soil pH and microbial biomass (Sebai et al., 2007). In our study, the intersample difference in pH was too small to support any correlation between the distribution of pH and mineralization potential, suggesting that pH does not play any role in the vertical variability in mineralization potential of phenoxy acid herbicides in the present aquifer sediment.

The small increase in 16S rRNA gene abundance in nonactive samples might be attributed to small disturbances of the subsurface sediment in the initialization of the experiment, for example, small changes in oxygen content. The bioavailability of other carbon sources than phenoxy acid herbicides in the sediment might increase 16S rRNA gene abundance, but not particularly tfdA genes. The lack of phenoxy acid mineralization activity also explains undetectable tfdA genes in nonactive samples.

The depth distribution of phenoxy acid mineralization correlated highly with tfdA gene copy number following mineralization, thus suggesting that the abundance of total tfdA genes particularly tfdA class I and class III genes – may determine the vertical variability in 2,4-D and MCPA mineralization potential. Moreover, tfdA gene diversity differed following exposure to the two compounds, with the number of tfdA class III genes being 100-fold greater and the number of tfdA class I genes being 10-fold greater following exposure to MCPA than to 2,4-D. This is in agreement with other studies showing that the tfdA class III gene is involved in the degradation of MCPA (Bælum et al., 2006) and mecoprop (Zakaria et al., 2007; Rodriguez-Cruz et al., 2009). The difference in tfdA gene copy number following mineralization of the two herbicides indicates growth of two different microbial populations, even though the molecular structure of the two herbicides is similar. The class I and III tfdA genes were found in β- and γ-proteobacteria and usually located on broad-host-range, self-transmissible plasmids (McGowan et al., 1998). However, the class II tfdA genes are less widely distributed and found only in the chromosomes of Burkholderia-like branches explaining the minor finding of class II tfdA genes in our aquifer sediments.

Centimetre-scale vertical variability in tfdA abundance has not previously been reported. It appears that degraders harbouring tfdA class III gene predominated in the samples able to mineralize MCPA at all depths in the sediment column. This could be attributable to indigenous populations having developed owing to continuous exposure to low concentrations of MCPA leaching from the overlying farmland. Similar adaptation of microbial communities has been reported following continuous exposure to low herbicide concentrations in a comparable aquifer (Lipthay et al., 2003), as well as in agricultural soils treated with 2,4-D and MCPA where the exposure level is higher (Bælum et al., 2008).

Conclusion

This study demonstrates the existence of centimetre-scale vertical variability in 2,4-D and MCPA mineralization potential of aquifer sediment below the groundwater table. Mineralization was fastest in sediment close to the groundwater table and exhibited considerable variability at increasing depth. The findings indicate that the vertical variability in 2,4-D and MCPA mineralization may be determined by the abundance of total tfdA, tfdA class I and class III genes. In sediment samples able to mineralize MCPA, the greater abundance of total tfdA and tfdA classes may be due to the presence of microbial degraders that have adapted to continuous exposure to MCPA. In addition to correlating with the abundance of catabolic genes, vertical variability in mineralization potential also correlated with the organic content of the sediment.

Groundwater systems are complex environments, and it is challenging to understand all interacting hydrological, geochemical and microbiological aspects. The elucidation of the factors controlling the small-scale variability in mineralization potential is very important to predict the fate and transport of phenoxy acid herbicides for groundwater management and protection. Our study underscores the necessity to take small-scale variability in mineralization into account when assessing the impact of pesticides on groundwater as the natural herbicide attenuation potential of these oligotrophic environments seems to be associated with highly localized microbial hotspots in the transition zones acting as biological filters.

Acknowledgements

This project was supported by Research Training for Good European Groundwater Resources, Support for training and career development of researchers (Marie Curie), Grant Agreement Number 212683. We thank Jens Bisgaard and Carsten Guvad (GEUS) for technical assistance.

Footnotes

  • Editor: Alfons Stams

References

View Abstract