Reductive dechlorination in recalcitrant sources of chloroethenes in the transition zone between aquifers and aquitards

In the transition zone between aquifers and basal aquitards, the perchloroethene pools at an early time in their evolution are more recalcitrant than those elsewhere in the aquifer. The aim of this study is to demonstrate that the biodegradation of chloroethenes from aged pools (i.e., pools after decades of continuous groundwater flushing and dissolution) of perchloroethene is favored in the transition zone. A field site was selected where an aged pool exists at the bottom of a transition zone. Two boreholes were drilled to obtain sediment and groundwater samples to perform chemical, isotopic, molecular, and clone library analyses and microcosm experiments. The main results were as follows: (i) the transition zone is characterized by a high microbial richness; (ii) reductively dechlorinating microorganisms are present and partial reductive dechlorination coexists with denitrification, Fe and Mn reduction, and sulfate reduction; (iii) reductively dechlorinating microorganisms were also present in the zone of the aged pool; (v) the high concentrations of perchloroethene in this zone resulted in a decrease in microbial richness; (vi) however, the presence of fermenting microorganisms supplying electrons for the reductively dechlorinating microorganisms prevented the reductive dechlorination to be inhibited. These findings suggest that biostimulation and/or bioaugmentation could be applied to promote complete reductive dechlorination and to enhance the dissolution of more nonaqueous phase liquids (DNAPL).


Introduction
Chloroethenes are chlorinated solvents that belong to the group of nonaqueous phase liquids (DNAPL) and have been involved in numerous contamination episodes. Mackay et al. ( 2006 ) compiled their chemical and physical properties. The latter facilitate its migration as free phase through the subsurface (Mercer andCohen 1990 ; Parker et al. 2003 ). Furthermore, the dissolved contaminant can penetrate lowconductivity materials by molecular diffusion (Parker et al. 1994; Chapman and Parker 2005; Sale et al. 2008. This process hinders the remediation of the aquifer because of slow back diffusion of the compounds from the lowconductivity material (Chapman and Parker 2005; Parker et al. 2008; Chapman et al. 2012. The transition zones between granular aquifers and basal aquitards were described by Parker et al. ( 2003 ) as a reasonable paradigm for the DNAPL source area architecture in granular aquifers. Such zones are located at the bottom of many aquifers and are characterized by the presence of numerous thin siltyclay layers interstratified with coarsergrained layers (i.e., sands and gravels), which results in a considerable heterogeneity. Furthermore, retardation factors and contaminant concentrations, in the case of aged sources, are higher in the finegrained layers (less conductive) than in the adjacent coarsergrained layers (Rivett et al. 2014 ). An aged pool is a former pool of free phase, which is currently an ancient pool where the DNAPL partially occupies the pores at saturations below the residual saturation value (Hartog et al. 2010 ) and constitutes an immobile phase retained by capillary forces. In addition, the groundwater flow is very slow in the transition zones because the hydraulic conductivity is lower than in the shallower part of the aquifer. For this reason, DNAPL is dissolved and flushed away from the more conductive zones of the aquifer after decades of groundwater flow when the source is aged, whereas DNAPL concentrations may remain close to or at solubility in the transition zones (Parker et al. 2003 ). This low contaminant mobility in transition zones should be pointed out since it implies that DNAPL sources in these zones are recalcitrant (much more than those in the rest of the aquifer), which has far reaching implications for the environment. Puigserver et al. ( 2013 ) demonstrated that numerous biogeochemical processes coexisted in the transition zone to a basal aquitard at a particular site in La Pineda (Camp de Tarragona, NE Spain). In this site, the aquifer was contaminated by chlorinated solvents (chloroform and carbon tetrachloride), and it was situated in a geological context of distal prograding alluvial fan deposits (sheet floods) formed by sand layers and interbedded silts rich in organic matter and with a large diversity of microorganisms. In the transition zone of this aquifer, layers with finegrained materials showed greater microbial richness than layers of coarsergrained materials. Groundwater flow through the more conductive layers supplied nutrients, contaminants, and electron donors, which led to the biodegradation of dissolved chlorinated solvents.
Chloroethenes may be recalcitrant under some conditions over long periods (several decades or longer). However, they can also be biodegraded, for example, under anoxic conditions by biotic reductive dechlorination (Wiedemeier et al. 1998; Bradley 2003, which is carried out by organohaliderespiring bacteria (OHRB). Moreover, some studies show that the presence of dechlorinating activity can significantly enhance the dissolution rate of the source of perchloroethene (PCE) McCarty 2000 , 2002 ). Reductive dechlorination of chloroethenes takes place by sequential dechlorination from PCE to trichloroethene (TCE), to 1,2cisdichloroethene (cDCE), which is the most common metabolite in TCE biodegradation (Bouwer 1994 ), to vinyl chloride (VC), and to ethene or ethane (Vogel et al. 1987 ; Tiehm andSchmidt 2011 ). Reductive dechlorination of PCE and TCE may take place under nitrate (van der Zaan et al. 2010 ), Mn, and Fereducing conditions (Bradley and Chapelle 2011 ), under sulfatereducing and methanogenic conditions (Bouwer 1994; Chapelle 1996; Bradley 2003, especially if an excess of electron donors is supplied to achieve substantial dechlorination (Aulenta et al. 2007 ). The reductive dechlorinating sequence may be wholly or partially inhibited by the competition for electron donors depending on the environmental conditions. This competition is between communities of OHRB and communities of anaerobic hydrogenotrophic (including reducers of NO , Mn , Fe , and SO ), autotrophic methanogenic, and homoacetogenic microorganisms (Wei and Finneran 2011 ). Among the microorganisms capable of reductively dechlorinating chloroethenes are the genera Desulfitobacterium, Clostridium, and those related to Dehalobacter restrictus (Miller et al. 1996; Kim et al. 2006. Dehalococcoides mccartyi strain 195 is the most widely 3 − 4+ 3 + 4 2− studied strain that dechlorinates PCE to ethene, whereas other OHRB are not usually able to dechlorinate PCE past cDCE and result in an accumulation of metabolites that are more toxic than PCE in the aquifer (MaymóGatell et al. 1997; Sung et al. 2003; Yoshida et al. 2007 ). Oxidizing microorganisms have also been identified, such as aerobes, which are able to oxidize chloroethenes to CO (Verce et al. 2002 ). High concentrations of chloroethenes in the source may inhibit microbial activity (National Research Council 1999), causing a decrease in the microbial richness of the population due to their toxicity (Haack and Bekins 2000 ).
Although Puigserver et al. ( 2013 ) demonstrated that numerous biogeochemical processes coexisted in the transition zone in the plume, a review of the literature reveals a significant gap in knowledge concerning the biogeochemical processes that take place in transition zones in which DNAPL sources are found. The present study aims to fill this gap by analyzing the reductive dechlorination occurring in this transition zone in a source area and to study how it is affected by other biogeochemical processes. To this end, a working hypothesis was formulated: the reductive dechlorination of chloroethenes in these transition zones occurs despite the potential inhibition that arises as a result of microbial competition and the partial inhibition of microbial communities caused by the high concentrations of PCE. To test this hypothesis, a field site on a confined aquifer consisting of Pliocene prograding alluvial fan deposits was selected in an industrial area in Vilafant (Alt Empordà, NE Spain), approximately 150 km to the north of Barcelona. A transition zone to a basal aquitard exists at the lower section of the aquifer between the depths of 5.60 and 7.50 m, where a PCE DNAPL source was detected by the Catalan Water Agency (ACA).

Site description
History of the site The site consists of a confined aquifer made up of Pliocene materials. Chloroethene contamination was detected at the site in 1980, but it is not known when this originated. The main contaminant is PCE, which was used as a degreaser of vehicle parts at a nearby industrial plant serving the automotive industry. PCE as DNAPL was most likely dumped into an old abandoned agricultural well located in the plant. Groundwater contamination resulted in concentrations of PCE that remained high in 2012, the maximum value being 104 μmol/L (i.e., 17,300 μg/L), whereas TCE and cDCE were 0.30 and 0.26 μmol/L, respectively (i.e., 39 and 25 μg/L, respectively). Except for monitoring natural attenuation, other remediation measures have not been implemented by the ACA.
The monitoring network consists of 12 conventional wells installed by the ACA. In addition, two boreholes (BF1UB and BF2UB) were drilled by our research group in January 2011 and were subsequently equipped as multilevel wells (F1UB and F2UB in Fig. 1 ; the construction characteristics of these wells are given in Table 1 ).

Fig. 1
Monitoring network of the site. PCE concentrations in the plume (2011)  Geological and hydrogeological framework The Pliocene prograding alluvial fan deposits, which belong to the geological context of the Pyrenees range and that constitute the subsurface materials at the site (IGME 1994; IGC 1996, are divided into three lithological units. The upper unit of the sequence is dominated by proximal deposits of gravels and sands that are mostly channelized and display high porosity despite the presence of some isolated discontinuous intervals of clays and sands. The middle unit consists of deposits of distal alluvial fans (sheet floods) made up of sand layers and interbedded silts that are rich in organic matter. The lower unit of the sequence corresponds to basin plain deposits that are mainly composed of fine sands and silts. Multilevel wells F1UB and F2UB and other conventional wells of the monitoring network reach the depth at which this lithological unit is located. The stratigraphic correlation of borehole logs showed that these units dip toward the northeast.
The aquifer is composed of two sections: (i) a part of the proximal coarse grained alluvial fan deposits (upper unit) in which hydraulic conductivities range from 10 to 20 m/day and (ii) the distal alluvial fan deposits (middle unit), whose hydraulic conductivities range between 1 and 10 m/day. Gravel paleochannels of the distal alluvial fan deposits act as drainage lines, coinciding with the general flow toward the northeast as shown in Fig. 1 . The average annual fluctuation of the piezometric level is 1.50 m, and the average horizontal hydraulic gradient is 0.020. Groundwater flow velocities range between 0.20 and 0.41 m/day in the upper unit and between 0.02 and 0.20 m/day in the middle unit. The fine sands and silts of the lowermost lithological unit constitute an aquitard.

Hydrostratigraphic units differentiated in the subsurface
Five hydrostratigraphic units of Pliocene age were differentiated from surface to bottom ( Fig. 2 ): (1) the unsaturated zone, which is made up of gravels and coarse, medium and fine sands; (2) the upper aquitard (upper discontinuous thin aquitard), which mostly consisted of clays and was crossed by subvertical microfractures; (3) the upper part of the aquifer was mainly constituted by gravels; and (4) the lower part of the aquifer, which comprised the transition zone to the basal aquitard, was made up of alternating gravels and coarse sands with numerous interbedded layers of medium to fine sands and silts at the centimeter to decimeter scale. The number of lithological and textural changes in this unit was among the highest in the geological profile (Table 2 (A)). A stratigraphic correlation with cores of conventional piezometers of the monitoring network showed that these layers had a limited lateral extension. (5) Lastly, the basal aquitard consisted of fine laminar sands and silts crossed by a dense network of subvertical microfractures. alternating gravels and medium to fine sands (gravels with no matrix and sands with a high siltyclayey matrix content). UDTA (3.30-4.37 m): greenish clays with interbedded fine and medium sands (a network of vertical microfractures intersects these materials). UPA (4.37-5.88 m): poorly sorted gravels and small proportion of coarse and medium sands (gravels and sands with a very little matrix content). Very few interbedded layers of silts. TZBA (5.88-7.50 m): alternating coarse to fine sands with numerous interbedded layers of silts (sands with very high content of siltyclayey matrix, especially when closer to the contact with the BA). BA (7.50-17.00 m): very fine sands and, in much lesser proportion, silts (all these materials are affected by vertical microfractures). BF2UB: USZ (0.00-3.58): alternating medium sands with interbedded gravel and coarse sand layers (gravels and sands with higher siltyclayey matrix content than in BF1UB). UDTA (3.58-4.18 m): greenish clays (intersected by vertical microfractures). UPA (4.18-5.60 m): alternating gravels and medium to fine sands (similar content of silty clayey matrix in gravels and sands than in the case of BF1UB). Very few interbedded layers of silts. TZBA (5.60-7.40 m): gravels and coarse sands with interbedded layers of medium to fine sands (gravels with no matrix and sands with lower siltyclayey matrix content than in BF1UB). BA (7.40-20.20 m): very fine sands with silty matrix (affected by vertical microfractures). Thin interbedded layers of medium sands to fine gravels Borehole drilling, geological core description, and installation of multilevel wells Boreholes BF1UB and BF2UB (17.00 and 20.20m depth, respectively, and 2.50 m from each other) were drilled by rotary drilling with a diamond crown bit, which had a provisional metal sleeve casing with an external diameter of 127 mm. A core sampler (85mm internal diameter) was used to recover the cores (37.7cm length on average). The geology of cores was described in detail at the site (Fig. 2 ), which allowed the differentiation of the coarsergrained layers from the finer layers, where a significant mass of chloroethenes tends to be stored. The drilling operations and the core sampler are described by Puigserver et al. ( 2013 ). The boreholes were equipped as multilevel wells (F1UB and F2UB, multilevel CMTtype 7 channels, Solinst) following the protocol established by Einarson and Cherry ( 2002 ) (see Table 1 for the construction characteristics of these multilevel wells). Of the seven ports of F1UB and F2UB, ports 1 and 2 were screened in the unsaturated zone. Port 7 in the multilevel well F1UB is at a depth of 6.91 m and is 7.20 m in F2UB (Table  1 ).

Core sampling, analytical determinations in porewater, and sediment and screening for DNAPL
In view of the considerable heterogeneity in the transition zone, intensive sampling of continuous cores from the two boreholes was essential to characterizing the presence of DNAPL, the vertical and lateral extent of DNAPL source, the distribution of DNAPL as residual or in pools, and the contaminant mass in porewater and sediments. Core sampling was also necessary: (1) to study the biogeochemical conditions under which the biodegradation processes occurred in the subsurface, (2) to analyze the depth variation of contaminants that resulted from degradation processes expressed as biodegradation haloes in the depth profile of concentrations (see " Reductive dechorination in the transition zone " section) and to verify the presence of microbial communities in sediment samples and to study their role in the degradation of chloroethenes.
Cores were exhaustively sampled in the field to characterize the vertical distribution of PCE, TCE, cDCE, transdichloroethene (tDCE), 1,1DCE, and VC in porewater. All of these compounds were analyzed at each sampling depth, with 26 and 37 samples in BF1UB and BF2UB, respectively. The percentage of organic carbon (f ) in sediments (26 and 32 samples in BF1UB and BF2UB, respectively), the total Fe and Mn sorbed in the fine fraction of sediments (12 and 32 samples, respectively), and the richness of microbial communities (see " Molecular analyses and clone library analyses " section) in these sediments (10 samples) were also analyzed. All samples for f , Fe, and Mn correspond to a oc oc sample of chloroethenes at the same depth. The distance between sampling points varied from 0.03 to 0.73 m (0.23 m on average) for chloroethenes, from 0.03 to 0.77 m (0.24 m on average) for f , from 0.07 to 1.02 m (0.32 m on average) for Fe and Mn, and from 0.21 to 2.16 m (0.77 m on average) for richness of microbial communities. The sampling procedures are described in Puigserver et al. ( 2013 ). The sampling procedure and conservation protocol for chloroethene analysis, as well as the calculations of porewater concentrations were an adaptation of the protocol followed by Parker et al. ( 2003 ) and Chapman and Parker ( 2005 ) (see Electronic Supplementary Material for more details on sampling procedures, sampling criteria, and number of replicates and field, transport, reactant and instrumentation blanks, and conservation protocols, as well as for sample pretreatments and analytical techniques). Sudan IV screening for the colorimetric determination of residual DNAPL or free phase was conducted while drilling the two boreholes following the method described by Parker et al. ( 2003 ) and Hartog et al. ( 2010 ).

Groundwater sampling and conservation protocols
Groundwater chemistry in the transition zone where DNAPL was present also provided the chemical characterization of the dissolved phase contaminants at the depths of the source of contamination. Groundwater was sampled using multilevel wells. Their sampling ports, located at different depths, provided groundwater samples that offered a highresolution image of the depth distribution of contaminants (Chapman et al. 2007). Changes in the concentrations of contaminants, oxidants, and metabolites were used to confirm the activity of different microorganisms and identify the processes involved. "In all cases, the sampling and conservation protocols reported by Puls and Barcelona ( 1996 ), Trevors ( 1996 ), Wiedemeier and Haas ( 2002 ), and Johnston ( 2006 ) were followed. Groundwater from ports 3 to 7 was sampled to analyze PCE; TCE; cDCE; tDCE1,2transdichloroethene (tDCE); 1,1DCE; VC; and the C fractionation of chloroethenes, nitrate, nitrite, sulfate, Fe, Mn, and dissolved oxygen (DO) in the aquifer (See Electronic Supplementary Material for more details on groundwater sampling procedures, conservation protocols, sample pretreatments, and analytical techniques).

Microcosm setup and monitoring
Two microcosm experiments were conducted. The first simulated the natural attenuation of PCE in the upper part of the aquifer, and the second simulated the natural attenuation of PCE in the transition zone to the basal aquitard. Each oc experiment consisted of two active tests (i.e., in which microorganisms were living) and two control tests (i.e., in which microorganisms were killed). Water samples from the two microcosm experiments were collected to study the time evolution of concentrations of the main inorganic electron acceptors (SO , NO , and NO ), PCE, TCE, isomers of DCE, VC, ethene, ethane, methane, Mn , and Fe (determinations of δ C values of chloroethenes or other organic compound in microcosm were not conducted). Sodium azide (N Na Fluka, purum pa) was added to the microcosm water samples immediately after being collected to inhibit bacterial activity. Before analysis, the vials containing the samples were stored in a cold chamber at 4 °C in total darkness. The temperature in the laboratory and inside the anaerobic chamber was maintained constant throughout the experiment (16 °C, corresponding to the annual average temperature of the aquifer) (see Electronic Supplementary Material for more details on microcosm setup).

Molecular analyses and clone library analyses
Core samples were collected on site, were stored in sterilized vases, and were frozen on site. Molecular analyses were performed to verify the presence of microbial communities in sediment samples and to study their role in the degradation of chloroethenes. The richness of microbial communities was also assessed. This parameter refers to the number of different species in a community. The number of sediment samples was 21 for molecular analysis using the terminal restriction fragment length polymorphism (TRFLP) technique and nested PCR, to identify Dehalococcoides (primer 582f, Duhamel et al. 2004 ; andprimer 728r, Löffler et al. 2000 ), Geobacter (primers Geo73f and Geo485r; Duhamel and Edwards 2006 ), and Dehalobacter (primers Deb179f and Deb1007r; Schlötelburg et al. 2002 ), which are genus with wellknown dechlorinating species. Replications were made (in duplicate) of each of the three restriction enzymes used. However, of the three assessed genus, only Dehalococcoides gave positive results since Dehalobacter and Geobacter were below the detection limit of the approach applied. As only Dehalococcoides spp. are known to be capable of the complete dehalogenation and detoxification of chloroethenes, this was one of the key groups of microorganisms to be considered. In addition, the clone libraries of two sediment samples taken in the transition zone were established. One of the samples originated at the depth where residual DNAPL was detected when drilling the boreholes (7.35 m deep) evidencing the presence of an aged pool (see " Reductive dechorination in the transition zone " section) and the other immediately above this pool (6.90 m deep). The analyses were performed at the Helmholtz Centre for Environmental ResearchUFZ (Leipzig, Germany). Genomic DNA was extracted from 1.10 g of sediment with NucleoSpin® Soil de Macherey & Nagel following the manufacturer's protocol to perform the TRFLP "and clone library analysis (see Electronic Supplementary Material for more details on molecular analyses and clone library analyses).

Results and discussion
Biogeochemical processes in porewater and groundwater Reductive dechorination in the upper part of the aquifer Pronounced peaks of PCE in porewater were located immediately below the contact upper aquitardupper part of the aquifer at depths of 4.80 m (6.94 μmol/L) and 4.51 m (2.15 μmol/L) for BF1UB and BF2UB, respectively (Fig. 3 a, labels 1A and 2A ). Although Sudan IV showed no signs of DNAPL currently present in the upper part of the aquifer, a fraction of the PCEDNAPL that previously penetrated into the upper aquitard through the vertical microfractures would have reached these depths in the upper part of the aquifer and remained as residual DNAPL (i.e., as trails of immobile discontinuous DNAPL droplets or ganglia) trapped interstitially in the gravels and sands (in a similar manner to that described by Sale et al. 2008 ) of the top of this unit (Fig.  2 ), where groundwater conditions were oxidizing (DO concentration of 8.00 mg/L) because of DO from the areas upgradient of the plant. Nitrification took place in F1UB (Fig. 4 a, port 3) owing to these oxidizing conditions, which are consistent with the low groundwater concentrations of TCE (0.010 and 0.011 μmol/L for the 2011 and 2012 surveys, respectively), suggesting a lower PCE degradation (with 2.06 and 0.48 μmol/L of PCE for the 2011 and 2012 surveys, respectively) than in the ports located in the transition zone (see " Reductive dechorination in the transition zone " section). The prevalence of the gravel and coarse sands in the upper part of the aquifer (Table 2 (B)) and the lower percentage of layers with siltyclayey matrix (Table 2 (C)) give this unit a more homogeneous character and higher hydraulic conductivity than those of the upper and lower stratigraphic units. These explain why PCEDNAPL pools did not accumulate and peaks of PCE did not form elsewhere in the upper part of the aquifer. A fraction of PCE residual DNAPL penetrated as dissolved phase into the fine matrix of sediments by molecular diffusion. Back diffusion from this matrix may cause the higher remnant groundwater concentrations in the upper part of the aquifer. Back diffusion from fine matrix also explains the lower groundwater concentrations in the central part of this unit, which was not as affected by DNAPL (Fig. 4 , port 3). Redox conditions continued to be oxidizing in this part of the hydrostratigraphic unit (the average DO concentration was 8.45 mg/L between 5.00 and 5.50m depth), which accounts for the absence of PCE reductive dechlorination as corroborated by the low concentrations of 3 4 metabolites in the porewater (the TCE, cDCE, and VC concentrations were below the LOQ, Fig. 3 a) and groundwater (Fig. 4 f, g; port 4). In addition to these low concentrations, the isotopic composition of PCE in groundwater at port 4 in F1UB (−25.04 ± 0.18 ‰) was the lightest in the whole profile, which is evidence that PCE degradation does not occur in this unit according to Aelion et al. ( 2009 ).  Since PCE degradation was very little in this unit at these depths, it is reasonable to suppose that, although residual DNAPL was not currently observed, the registered molar concentrations of PCE and TCE in porewater represent an estimation of the initial molar composition of the DNAPL that penetrated (assuming that DNAPL was formed by the mixture of these two compounds). However, this is only an approximation because it is impossible to know the exact effect of the changing solubility of these two compounds in the resulting mixture. Thus, using the average molar concentrations between depths of 3.45 and 4.80 m of PCE and TCE (other chloroethenes were not detected at these depths), i.e., 5.95 and 0.052 μmol/L, respectively, molar fractions of 99.12 and 0.88 %, respectively, were obtained. The low molar fraction of TCE is evidence that this compound occurred as an impurity in the initial DNAPL (hence, neglecting the effect of the changing solubility most likely did not significantly influence the result of the calculation).

Reductive dechorination in the transition zone
Reducing conditions prevailed throughout the year in the groundwater in this unit (DO varied between 1.25 and 0.89 mg/L), giving rise to denitrification with increasing depth as evidenced by the decrease in nitrate and the formation of nitrite from port 5 to 7 in F1UB and from 4 to 7 in F2UB (Fig. 4 a, b). Furthermore, reduction processes of Mn were identified in sediments at a depth of 5.69 m in BF2UB (Fig. 5 b, label 2b ) accompanied by an increase in Mn in the groundwater from these ports (Fig. 4 d). In addition, the Sudan IV test conducted during the drilling of the two boreholes evidenced the current presence of residual DNAPL on the geological contact with the basal aquitard (see green arrows on Fig. 3 a), which accounts for the two large peaks of PCE detected near this geological contact in BF1UB (112.86 μmol/L at a 7.35m depth, Fig. 3 a, label 1A ) and BF2UB (8.87 μmol/L at a 7.27m depth, Fig. 3   The reduction of sulfate was observed in ports 6 and 7 (Fig. 4 c), which suggests that the conditions (especially in port 7) were sulfate reducing. These conditions were created as a result of a lower input of DO into the source and the consumption of this oxygen in the oxidation of the solid particulate and dissolved organic matter in the source. Hydraulic conductivity decreased because a part of the porosity was occupied by the residual DNAPL (Fetter 1993 ). This decrease resulted in a smaller flow velocity in the source and therefore in the aforementioned lower input of DO. In this zone near the contact with the basal aquitard, the presence of a steep concentration gradient of Fe (with a maximum of 667.0 μmol/g; Fig. 5 a, label 1A ) and another of Mn (33.9 μmol/g; Fig. 5 b, label 1B ) in the sediments of BF1UB at a 7.35m depth is consistent with the aforementioned sulfatereducing conditions. Thus, watersediment equilibrium is achieved at this depth in which a fraction of the dissolved Fe and Mn precipitates as sulfide and/or carbonate minerals under reducing conditions (Garrels andChrist 1965 ; Brookings 1988 ).
The numerous textural changes (Table 2 (A)), the fact that the number of thin layers of fine materials exceeded that of coarsegrained materials in the transition zone (Table 2 (B)), and the scarcity of sand and gravel layers with no matrix (Table 2 (C)) demonstrate that the medium in this zone is heterogeneous. Despite this heterogeneity, the detailed description of the transition zone of the boreholes BF1UB and BF2UB in Fig. 2 is perfectly embodied in the definition of transition zone (see " Introduction " section). This heterogeneity led to the accumulation of a pool of PCE on the geological contact transition zonebasal 1 1 aquitard in the past (7.50m depth in F1UB and 7.40m depth in F2UB). Concentrations of the two large peaks of PCE detected near the geological contact with the basal aquitard in BF1UB and BF2UB, which, in milligram per liter, are 14.8 and 1.2 mg/L, respectively, were lower than the current PCE effective solubility value in the residual DNAPL (208.2 mg/L at 25 °C). Given that DNAPL was detected, these lower concentrations than the effective solubility are consistent with the presence of residual PCEDNAPL (Feenstra et al. 1991 ) at DNAPL saturation lower than the residual saturation (Parker et al. 2004 ) (see dashed red line in Fig. 3 a). Therefore, the aforementioned pool is currently an aged pool (i.e., composed of remnants of PCE free phase found in an immobile residual DNAPL form; Hartog et al. 2010  Furthermore, the predominance of lower hydraulic conductivity layers in the transition zone resulted in a low groundwater velocity that hindered the dissolution of the PCE pool. The subsequent lower rate of dissolution contributed to the recalcitrant character of PCE in this unit (Parker et al. 2003; Chapman and Parker 2005; Adamson et al. 2014, which contrasts with the upper part of the aquifer (where only few interbedded lowconductivity layers exist, and hence, groundwater velocity was greater; Fig. 2 and Table 2 (B)). Moreover, the storage of PCE by molecular diffusion (Cherry et al. 2006) in the finegrained sediments of the transition zone contributed to the persistence of PCE in this unit, as evidenced by the progressive PCE increase in the porewater of the two boreholes (Fig. 3 a, 1A and 2A ).

AQ4
As for the depth evolution of chloroethene concentrations, the existence of two depths with steep concentration gradients in TCE in the porewater of fine grained layers at depths of 6.36 and 6.90 m in BF1UB, with maxima values of 10.08 and 7.29 μmol/L, respectively (Fig. 3 b, labels 1B2, 1B3), related to a maximum in cDCE (1.05 μmol/L) and VC (0.26 μmol/L) was noteworthy (neither of the latter two compounds is shown in Fig. 3 ). These four maxima of concentration were centered on a PCE minimum at a depth of 6.71 m (0.03 μmol/L; Fig. 3 a, label 1a ) and constituted a biodegradation halo (halo 2), which is similar to another halo in the unsaturated zone (halo 1; Fig. 3 a, b), indicating the biotic reductive dechlorination of PCE and metabolites, as described by Bradley ( 2003 ) and Puigserver et al. ( 2013 ). According to the observations of the latter authors, a biodegradation halo of parent and metabolite compounds is a depth interval in the profile of concentrations where a gradual increase in a metabolite (e.g., TCE) reaches a maximum and then progressively decreases, which results in a steep concentration gradient of this compound. An opposite variation of the parental compound (e.g., PCE) is produced parallel to the evolution of this metabolite. This also forms a steep concentration gradient.
In addition, the PCE groundwater at port 7 of the multilevel well F1UB (located at a depth of 6.91 m, i.e., at the zone of halo 2) showed isotopic fractionation, with a δ C value of −23.53 ± 0.07 ‰. This value corresponded to a Δδ of +1.5 ‰ with respect to the upper part of the aquifer (where no degradation of PCE was detected; see " Reductive dechorination in the upper part of the aquifer " section). This isotopic shift is compatible with the Rayleigh fractionation curves of PCE by reductive dechlorination with enrichment factors ε ranging from −0.4 to −16.4 ‰ (Hunkeler and Morasch 2010 ), which is more evidence of PCE biodegradation forming TCE. Moreover, the δ C value of the TCE in groundwater from port 7 of F1UB (−22.10 ± 0.30 ‰) was heavier than that of the parent PCE, suggesting that a fraction of the TCE was also biodegraded by reductive dechlorination (under sulfatereducing conditions) to form cDCE; therefore, the isotopic fractionation of TCE also occurred, according to Hunkeler et al. ( 2008 ). As for F2UB, isotopic fractionation of groundwater PCE was apparently not observed given a δ C value in port 7 of −25.02 ± 0.16 (at a 7.20m depth), which is a similar value to that obtained in the upper part of the aquifer (where reductive dechlorination did not occur, and where a δ C value of −25.04 ± 0.18 ‰ was recorded, see " Reductive dechorination in the upper part of the aquifer " section). However, the fact that the reductive dechlorination of PCE to cDCE occurred in some depth intervals provides evidence that lack of isotopic fractionation of PCE must likely due to the continuous dissolution of DNAPL at this depth masking any fractionation effect (Braeckevelt et al. 2012 ). The process of reductive dechlorination is also consistent with the occurrence of oxidation of solid organic matter, which is evidenced by the minimum f value recorded at this depth in BF1UB (0.0014 % at 7.35 m; Fig. 5 c label 1c ) and in BF2UB (0.0069 % at 7.03 m; Fig. 5 c) (USEPA 1998; AzadpourKeeley et al. 2001. Although the aged pool has led to an increase in the PCE concentration in the groundwater in port 7 of the two multilevel wells and despite that the biogeochemical processes that take place compete with one another (i.e., 4 13 13 13 13 oc 4 reductive dechlorination, denitrification, reduction of Fe and Mn, and sulfate reduction), it is noteworthy that these circumstances do not completely inhibit the reductive dechlorination of PCE to cDCE because steep increases from port 6 to port 7 in TCE and cDCE were recorded at both boreholes (Fig. 4 f, g). The predominance of metabolites such as TCE and cDCE in the groundwater and porewater suggests that the reductive dechlorination sequence was not completed, according to Bradley ( 2011 ) andMaymóGatell et al. ( 2001 ), although low VC concentrations in port 7 of F1UB were also detected, with concentrations of approximately 0.080 μmol/L (not shown in Fig. 4 ).
Furthermore, the occurrence of dissolved iron in groundwater of port 7 in both multilevel wells (Fig. 4 d) and also in the sediments of BF1UB in the transition zone (Fig. 5 a, label 1 A1) suggests the presence of iron minerals (Fig. 4 c, d), as mentioned above. For this reason, it cannot be ruled out that part of the dechlorination of PCE can take place abiotically in the presence of these ferrous iron minerals.

Evidence of reductive dechlorination from a microcosm experiment
The absence of reductive dechlorination on a field scale in the upper part of the aquifer contrasts with the incomplete reductive dechlorination sequence observed in the transition zone. A similar observation was made at the laboratory scale from the microcosm experiment because no significant variations were recorded in the PCE, TCE, and cDCE concentrations from the microcosm experiment after 267 days of simulating the in situ natural attenuation in the upper part of the aquifer (first experiment, Fig. 6 a). Conversely, a progressive decrease in the PCE concentration, accompanied by the formation of TCE, cDCE, and 1,1DCE, was recorded in the natural attenuation simulation experiment in the transition zone from day 106 (second experiment, Fig. 6 b) as a result of reductive dechlorination, according to Bradley ( 2003 ). The reductive dechlorination in this experiment became significant when the sulfate concentration declined from 43.73 mg/L (day 106) to 16.72 mg/L (day 267), i.e., after the sulfate reduction conditions were reached. Therefore, although incomplete, reductive dechlorination is a more important biogeochemical process in the transition zone than in the upper part of the aquifer, which supports the observations at the field site. Furthermore, controls of these two experiments did not show any variation in concentrations of the PCE injected or metabolite formation (Fig. 6 c, d). Therefore, abiotic degradation did not occur in the microcosm experiments, which does not imply that it does not occur at field scale, as indicated at the end of " Reductive dechorination in the transition zone " section.

Fig. 6
Aqueous concentrations in microcosm experiments used to simulate natural attenuation. Error bars show the range of variation in duplicates. a In the upper part of the aquifer (first experiment, active tests). Red line indicates that up to 267 days (end of experiment), concentrations did not substantially change. b In the transition zone (second experiment, active tests). Red line indicates that from day 106, there were substantial changes in concentrations. c In the upper part of the aquifer (first experiment, control tests). d In the transition zone (second experiment, control tests). TCE and cDCE concentrations of first sampling day in control tests correspond to the groundwater sample used in the experiments after purging (see " Microcosm setup and monitoring " section)

Effects of high concentrations of chloroethenes on microbial communities
The peaks of PCE in porewater immediately below the contact unsaturated zone upper aquitard (Fig. 3 a, labels 1A and 2A ; see " Reductive dechorination in the upper part of the aquifer " section) coincided with a decrease in richness, i.e., the number of different species in the community (Fig. 3 c, labels 1c2 and 2c1) when compared with the upper unit. This decrease is attributable to an increase in the specialization of microbial populations because many of these cannot adapt to high concentrations of PCE (Sleep et al. 2006 ). Moreover, the minimum value of PCE in porewater at a depth of 3.96 m in fine and medium sands within the upper aquitard at BF1UB (0.17 μmol/L, Fig. 3 a, label 1a ) coincided with a minimum of richness (Fig. 3 c, label 1c ). This minimum is related to the absence of groundwater flow in these sands, which prevents microbial communities from gaining access to the nutrients, electron donors, carbon sources, and growth factors (natural substances that stimulate the growth, proliferation, differentiation, and cellular healing) that they need. Therefore, the most suitable conditions for the development of microbial communities were not met in these sands. The concentrations of PCE in porewater recorded in the aforementioned pronounced peaks immediately below the contact upper aquitardupper part of the aquifer (Fig. 3 a, labels 1A and 2A ; see " Reductive dechorination in the upper part of the aquifer " section) corresponded with lower richness of microbial communities (Fig. 3 c; labels 1C and 2c for BF1UB and BF2UB, respectively) than those on the contact unsaturated zoneupper aquitard (Fig. 3 c).
In the biodegradation halo centered on a PCE minimum in porewater at a depth of 6.71 m (0.03 μmol/L; Fig. 3 a, label 1a ; halo 2, see " Reductive dechorination in the transition zone " section), a significant increase in the microbial richness of microbial communities was recorded in BF1UB at a depth of 6.90 m (Fig. 3 c label 1C , with maximum relative value of 9) compared with the top of the transition zone and the upper part of the aquifer. This increase is consistent with the coexistence of the aforementioned biogeochemical processes (denitrification, reduction of Fe and Mn, sulfate reduction, and reductive dechlorination; see " Reductive dechorination in the transition zone " section). The aforementioned high peaks of PCE in porewater in the transition zone near the contact with the basal aquitard (Fig. 3 a, label 1A and 2A ; in BF1UB and BF2UB, respectively) diminished the richness (Fig. 3 c, labels 1c and 2c ; with relative average values of 8 and 6 in BF1UB and BF2UB, respectively) compared with the upper part of the transition zone in contact with the upper part of the aquifer (Fig. 3 c, labels 1C and 2C in BF1UB and BF2UB, respectively). These findings would suggest that high concentrations of PCE negatively affect this parameter of the structure of microbial communities, as reported by Haack and Bekins ( 2000 ).

Dechlorinating community in the sediments of the transition zone
The nested PCR technique allowed the identification of Dehalococcoides sp. in the transition zone, and the clone library analysis allowed the identification of other microorganisms in this zone. The clone library was established to compare the microbial communities in the aged pool and halo 2 (Fig. 7 ). Particular attention was paid to the clones related to the microorganisms that are reductively or oxidatively dechlorinating and those that are fermenting (because of their relationship with reductively dechlorinating microorganisms).

Fig. 7
Phylogenetic tree of the identified bacteria in the two sampled depths of borehole BF1UB. This tree has been produced based on the partial sequences of the 16S rRNA. The total number of sequenced clones was 47 for each of the two samples. Blue: the sample taken at 6.90m depth at the maximum TCE concentration located in the biodegradation halo 2. Orange: the sample taken at 7.35m depth in the aged pool of PCE. Dechlorinating and fermenting microorganisms were identified at 6.90m depth, coinciding with the maximum of TCE concentration in porewater located in the biodegradation halo 2 (Fig. 3 b label 1B ), i.e., immediately above the aged pool of PCE. Among the fermenting microorganisms, uncultured Bacteroidetes, Propionibacterium acnes, and Pelosinus propionicus were identified. These microorganisms generate propionic acid and H (Patil et al. 2014 ). In addition, P. propionicus is capable of synthesizing the corrinoids that some OHRB of cDCE and VC need to carry out the complete reductive dechlorination of chloroethenes (Men et al. 2014  The greater richness of microbial communities, especially in BF1UB, coincides with the presence of fermenting (P. propionicus and Bacteroides), reductively dechlorinating (Dehalococcoides sp. and P. acnes), and oxidatively dechlorinating (D. acidovorans, A. denitrificans, A. junii, and C. basilensis) microorganisms. The presence of these microorganisms under conditions ranging from microaerophilic to the reduction of Mn, along with the coexistence of numerous biogeochemical processes (see " Reductive dechorination in the transition zone " section), is evidence of partially inhibited reductive dechlorination by microbial competition. However, the presence of OHRB and PCE metabolites and the fact that dissolved PCE presents isotopic fractionation (see " Reductive dechorination in the transition zone " section) show the existence of reductive dechlorination, which corroborates the working hypothesis. Furthermore, the presence of TCE and cDCE oxidatively dechlorinating microorganisms suggests that a fraction of these two compounds may have been degraded by a metabolic pathway different from that of reductive dechlorination.
As for the zone where the aged pool is located (at a depth of 7.40 m, on the geological contact transition zonebasal aquitard), the greater clone frequencies corresponded to those of microorganisms with the fermentative metabolism of the phyla Bacteroidetes (uncultured Bacteroidetes) and Firmicutes (P. propionicus, Clostridium sp., and uncultured Peptococcaceae) (Patil et al. 2014  In this geological contact, the high concentrations of PCE caused by the presence of an aged PCE pool partially inhibit the microbial communities, as suggested by the decline in the richness of microbial communities (Fig. 3 c label 1c ). However, a higher proportion of fermenting microorganisms (Bacteroides, Clostridium sp., Peptococcaceae, and P. propionicus) compared with the zone of halo 2 ensures an important supply of H as a source of electrons for the OHRB (Dehalococcoides and Clostridium), which have been identified in environment in the presence of DNAPL (Chang et al. 2000; Schaefer et al. 2010 ). In addition, the prevalence of sulfatereducing conditions favors the reductive dechlorination of PCE and TCE, as evidenced by the presence of cDCE and traces of VC in the BF1UB porewater, also corroborating the working hypothesis at this depth. Nevertheless, the identification of oxidatively dechlorinating microorganisms also at this depth suggests that conditions may be locally microaerophylic and that it cannot be ruled out that a fraction of TCE and cDCE may have been degraded by a metabolic pathway different from that of reductive dechlorination.
The determined sequence data of 16S rRNA genes of the bacteria in Fig. 7 have been deposited in the European Nucleotide Archive (ENA). The accession numbers are listed in Table 3 . Evidence that reductive dechlorination of PCE and TCE is favored in the transition zone in contrast to the upper part of the aquifer and implications for remediation Several pieces of evidence support the observation that biotic reductive dechlorination of PCE and TCE is favored in the transition zone: (1) the numerous geological heterogeneities and textural changes in the transition zone (see " Hydrostratigraphic units differentiated in the subsurface " section) enable the different microorganisms to gain access to nutrients, electron acceptors, electron donors, carbon sources, and growth factors reaching this unit along the layers of larger grain size, and hence pore size, materials. Microorganisms also have access to the solid organic matter and chloroethenes along the surface of contact with fine materials similar to that described by Puigserver et al. ( 2014 ) in an alluvial aquifer. (2) The transition zone is located at the interface between the oxic medium of the upper part of the aquifer (see " Reductive dechorination in the upper part of the aquifer " section) and the anoxic medium generated by the presence of the aged PCE pool at the transition zonebasal aquitard geological contact (see " Reductive dechorination in the transition zone " section), which resulted in a high microbial richness in this unit (Fig. 3 c, labels 1C and 2C ). 3) Numerous biogeochemical processes, such as denitrification, the reduction of Fe and Mn, sulfate reduction, and the reductive dechlorination of PCE and TCE (see " Reductive dechorination in the transition zone " section), related to a large richness of microbial communities (Fig. 3 c, labels 1C and 2C ) that includes the presence of OHRB (see " Dechlorinating community in the sediments of the transition zone " section), coexist in this zone (see " Effects of high concentrations of chloroethenes on microbial communities " section).
In contrast to the upper part of the aquifer (where the biodegradation of PCE and TCE was not observed), the transition zone groundwater showed presence of the metabolites of PCE and TCE and the isotopic fractionation of PCE (see " Reductive dechorination in the transition zone " section) and, hence, reveal the presence of reductive dechlorination.
Despite the presence of microorganisms capable of completing reductive dechlorination (Dehalococcoides and Clostridium), this process is not significant, and major amounts of TCE and cDCE accumulate in the transition zone. Redox conditions were not a limiting factor for complete reductive dechlorination because this process also did not occur in the transition zone 4 2 4 2 microcosm experiment, where sulfatereducing conditions were achieved (see " Evidence of reductive dechlorination from a microcosm experiment " section). These findings, and the fact that no significant variations in the PCE, TCE, and cDCE concentrations were recorded in the upper part of the aquifer natural attenuation microcosm experiment, suggest that one of the major limiting factors for completing reductive dechlorination may be a small supply of bioavailable electron donors. According to Rifai et al. ( 2011 ), this suggests that an adequate supply of fermentable substrates is needed for the production of dissolved hydrogen, which is used directly by OHRB as the electron donor. For this reason, biostimulation (and/or bioaugmentation) in the transition zone by adding electron donors could be more efficient than in the upper part of the aquifer, where the richness of microbial communities tends to be lower. Moreover, although microcosms were not replenished with electron donors and substrates, as would occur in the field, and despite the fact that this microcosm study is a snapshot in time, it reinforces the possible use of biostimulation and/or bioaugmentation to improve degradation rates in order to accelerate degradation and to enhance the subsequent dissolution of more DNAPL, which provides greater value and significance to our findings.
Moreover, we previously studied a contaminated transition zone by chlorinated solvents at another site (Puigserver et al. 2013 , see " Introduction " section). This transition zone was also characterized by a large diversity of microorganisms, which included OHRB (Puigserver et al. 2016), and also was located in deposits of distal prograding alluvial fans in the geological context of the Catalan Coastal Ranges (i.e., different to the context of the present study). Given that similar observations to those obtained in the current study were made, it is reasonable to think that in cases of transition zones contaminated by chlorinated solvents and geologically located in alluvial fan deposits similar to that of the current study, the characteristics of this contamination and its temporal evolution could be similar to that of the present work.

Conclusions
Our study shows the enormous variability in geochemical and microbiological results. In order to understand what this involves and in order to interpret this variability, it is essential to carry out a detailed geochemical and microbiological sampling. This variability is associated with (i) the heterogeneity of sediments (variations in hydraulic conductivity), (ii) the redox conditions, and (iii) the chloroethene concentration; i.e., it increases on highlow hydraulic conductivity boundaries, where DNAPLmass persistence occurs.
Furthermore, natural attenuation by biodegradation in porewater is noteworthy since it helps to maintain steep chloroethene concentration gradients (strong concentration variability) that create a large mass diffusion flux. This mass biodegrades at a rate that maintains these gradient steeps. The natural dechlorinating activity by OHRB and the more efficient natural transformation of PCE into TCE in the transition zone than in the rest of the aquifer are of considerable environmental significance. Therefore, bioremediation via reductive dehalogenation in the transition zone may be applied more efficiently to promote complete reductive dechlorination and to enhance the subsequent dissolution of more DNAPL in these zones than in other parts of the aquifer. Hence, removal of the source would be accelerated and the effects of back diffusion could be reduced or eliminated. Thus, biostimulation of the dechlorinating indigenous microbial communities and/or bioaugmentation may be conducted by making the site more anoxic and providing an electron donor.
Furthermore, the environmental importance of our work lies in the fact that many supply wells in the world exploit aquifers, which are constituted by materials that correspond to recent unconsolidated alluvial fan deposits and that form the principal groundwater source. Extraction wells are easy to drill and their groundwater is easily accessible. Alluvial fans fill basins where land use devoted to drinking water and to industry could lead to the contamination of aquifers with chlorinated solvents.