Skip to main content

BIOTREATMENT OF DRINKING WATER RESOURCES POLLUTED BY PESTICIDES, PHARMACEUTICALS AND OTHER MICROPOLLUTANTS

Final Report Summary - BIOTREAT (BIOTREATMENT OF DRINKING WATER RESOURCES POLLUTED BY PESTICIDES, PHARMACEUTICALS AND OTHER MICROPOLLUTANTS)

Executive Summary:
BIOTREAT is an EU financed research project (FP7) that started 1st January 2011 with 11 project partners including research institutions, waterworks and small and medium-sized enterprises (SME’s). The primary aim of BIOTREAT was to develop new technologies for bioremediation of drinking water resources contaminated with micropollutants such as pesticides and pharma-ceuticals. The technologies are based on the introduction and efficient exploitation of degrader bacteria into for example existing sand filters at waterworks. The project has been structured in 8 well-integrated work packages (WPs) each headed by a WP leader.

In WP1 new isolation techniques were developed and degrading bacteria were selected to be used in the other WPs. Furthermore, techniques for immobilisation of degrader bacteria on specific carriers was developed improving degradation efficiency and preventing loss of the bacteria from the treatment filters. WP2 focused on degradation kinetics and pathways at rele-vant micropollutant concentrations being in the ng to µg/l range. A method for measuring as-similabe organic carbon (AOC) was developed and it was shown that the content of AOC in the water was of decisive importance influencing degradation efficiency. WP3 examined the sur-rounding environment of the systems, focusing on the effect of sand filter ecology on intro-duced degrader bacteria. Tools to analyse molecular fingerprinting results in order to provide ecological parameters including richness (Rr), dynamics (Dy) and functional organization (Fo) were developed. Protozoa naturally living within waterworks sand filters was shown to predate on the degrader bacteria added, thereby limiting contaminant degradation. In WP4 knowledge gained in WP1-3 was integrated in mathematical modelling to improve the technology predict-ability and our understanding of the degradation process in general. This allowed the design of in situ and in reactor applications of microbial remediation necessary for the large scale applications carried out in WP5. Models were developed for both metabolic primary pollutant removal and cometabolic oxidation of micropollutant by ammonium oxidizing bacteria (AOB). WP5 was devoted to large-scale application of the developed metabolic and cometabolic re-mediation technologies. Up to 75% removal of 0.2 µg/l 2,6 dichlorobenzamide (BAM) was achieved in sand filters inoculated with Aminobacter sp. MSH1, degrading BAM metabolically. Immobilisation of MSH1 improved degradation efficiency for a longer period as it diminished loss of MSH1 from the filters. The cometabolic strategy, using ammonium oxidizing bacteria (AOB) was less promising at large-scale conditions probably due to the short hydraulic re-tention time of the filter. WP6 analysed the regulatory, safety, financial and environmental aspects of the developed technologies using cost-benefit analysis (CBA) and life cycle impact assessment (LCIA) approaches. The CBA and LCIA showed that the BIOTREAT metabolic strategy was competitive to granular activated carbon treatment being the most likely competitor technology to the BIOTREAT technology.

The main outcome of BIOTREAT has been disseminated to a broader audience of stakeholders, managers, expert and scientists within the water supply at the BIOTREAT open-end users meet-ings and at international conferences, symposia and workshops. BIOTREAT has contributed to the education of a new generation of environmental scientists as several early stage research-ers have been educated with funding from BIOTREAT including both PhDs and postdoc fellows. More information about the project can be found at the BIOTREAT homepage (www.biotreat-eu.org).

An extended version of this report which also contains figures and tables are presented in de-liverable 8.16 of the project.

Project Context and Objectives:
Millions of tonnes of organic xenobiotics are used each year worldwide, for example the pesti-cides used in agricultural production and applied to paved urban areas, along railways and roads, and to farmyards. Pharmaceuticals are considered to be emerging contaminants that enter the environment via the application of manure and sewage sludge to farmland, via efflu-ent from wastewater treatment plants and via accidental spillage during various industrial ap-plications. As a result of these extensive environmental inputs, European water bodies such as rivers, lakes and groundwater that are used as drinking water resources have become contami-nated with a wide range of organic micropollutants. There is, however, a striking contrast be-tween the input levels (up to several kg per hectare for pesticides) and the contaminant con-centrations detected in water bodies, which are normally in the microgram to nanogram per litre range (Schwarzenbach et al., 2006). The EU limit value for individual pesticides in drinking water is 0.1 µg/l, while that for multiple pesticides is 0.5 µg/l (Directive 98/83/EC). These con-centrations are very low, and many water bodies have therefore had to be abandoned as drink-ing water resources (Stockmarr, 2005).

The incredible capacity of soil and subsurface microbial communities to degrade a wide range of xenobiotic compounds is well known. In most cases where microorganisms are used for bio-remediation, however, the biodegradation processes operate as “black box” systems, and de-tailed knowledge of the microorganisms and the metabolic processes and pathways involved is lacking. As complete mineralisation of organic micropollutants may require specialised strains that are rarely present in the indigenous microbial community (e.g. Topp et al., 2000; Sørensen et al., 2007), bioaugmentation with degradative bacteria has been proposed as a strategy for remediation of contaminated drinking water resources. It is believed that bioaugmentation has a great but unexplored potential for drinking water remediation as decontamination of the wa-ter will take place in contained controlled-flow systems such as those used for conventional wa-ter treatment (e.g. sand filters and activated C filters). Development of successful bioremedia-tion-based strategies for treating drinking water resources is in its infancy, however, and the low concentration at which micropollutants are generally present in freshwater environments adds to the challenge.

The overall objective of BIOTREAT was to develop new technologies for bioremediation of drinking water resources contaminated with micropollutants such as pesticides and pharma-ceuticals. The technologies are based on the introduction and efficient exploitation of degrader bacteria in existing sand filters at waterworks. Once developed the technologies may also be transferred to other types of submerged biofilter systems such as mobile biofilters placed close to groundwater abstraction wells, sand barriers between surface waters and abstraction wells and subterranean protective barriers established to prevent micropollutants from entering into aquifers. To ensure efficient and reliable pollution degradation, the BIOTREAT technologies will make use of bacteria degrading target pollutants by either metabolic or cometabolic degrada-tion pathways. The availability of suitable degrader bacteria and the mixture of micropollutants to be targeted in the individual applications will determine the choice of removal strategy. A strategy based on metabolic processes will be exploited for micropollutants where metabolically degrading bacteria are available. For target pollutants where specific degrading bacteria have not been described, an alternative strategy relying on unspecific cometabolic degradation by methane or ammonium oxidising bacteria will be exploited.

The BIOTREAT project has been structured into 8 well-integrated work packages, each headed by a WP leader. WP1, WP2, and WP3 aimed at exploring the “black box” of bioaugmentation, providing in-depth knowledge about degradation, growth and survival of microorganisms in-troduced into sand filters. Together WP1-3 brought progress to front-line research in bioaug-mentation technology which in the subsequent WPs was translated into novel applied technol-ogies for water treatment.

The focus of WP1 was on the microorganisms, their physiology and their interactions with the matrix environment. The aim was to characterise existing metabolically and cometabolically pollutant-degrading bacteria regarding degradation of micropollutants at low concentrations and to exploit new isolation and enrichment strategies for degrader bacteria adapted to degra-dation at low contaminant concentrations. New micropollutant-degrading bacteria were isolat-ed and characterized in details. Based on the research degrader bacteria were selected to be used within other WPs. Furthermore, it was an aim to develop new carrier and encapsulation technologies for introducing degrader bacteria into biofilters/barriers ensuring their long-term survival and activity.

WP 2 emphasised the degradation processes, providing insight into metabolic pathways and metabolite formation. The aim was to study microbial growth and degradation kinetics at low contaminant concentrations and to identify rate-limiting factors for growth in waterworks sand filters. It was also an aim to identify contaminant degradation pathways and to elucidate the metabolite pattern for both metabolic and cometabolic degradation processes. Environmental factors being bottlenecks for degradation were also identified, including the availability of as-similabe organic carbon shown to limit growth and degradation in waterworks sand filters.

WP3 examines the surrounding environment of the systems, focussing on the effect of sand filter ecology on introduced degrader organisms. The aim was to study how the dynamics of microbial community structure influences the degradation of contaminants and how this can be controlled to increase the predictability of the degradation process. In addition it was an aim to study the effect of protozoan predation of degraders on degradation efficiency in drinking wa-ter facilities. To achieve the aims tools for interpretation of molecular fingerprinting pattern, including the range-weighted richness (Rr), dynamics (Dy) and functional organization (Fo) pa-rameters for the different experimental setups were developed.

The focus on WP4 was on models and predictability, where the knowledge gained in WP1-3 was integrated in mathematical modelling to improve technological predictability. The aim was to develop generalised models for both primary pollutant removal (metabolic) and cometabolic transformation. Low contaminant concentrations being in the microgram/l range typically found in polluted drinking water was emphasised in the modelling.

WP 5 focused on application and reliability through large-scale applications of the developed technologies, including laboratory-scale reactors and column experiments carried out in the laboratory or at waterworks. The aim was to transfer the developed metabolic and cometabolic bioaugmentation technologies from laboratory-scale experiments to submerged biofilter sys-tems at a waterworks and finally to field-scale implementation. The up-scaling effect on the performance of the remediation technology was determined and the system was further ad-justed to improve degradation efficiency.

WP6 contained an integrated assessment and performance validation of the developed tech-nologies. The regulatory, safety, financial and environmental aspects of the technologies were analysed using a life cycle impact assessment (LCIA) approach as well as cost benefit analyses (CBA). The aim was to use the LCIA and CBA to compare the BIOTREAT technologies to other technologies including granular activated carbon (GAC) filtration to help decision makers to select the best strategy for remediation of contaminated drinking water resources.

WP7 focused on dissemination of the project results, further exploitation of the technologies and the development and continuous improvement of the technology action plan containing the overall exploitation strategy.

WP8 was devoted to project management.

Project Results:
WP1 ORGANISMS – IDENTIFICATION AND CHARACTERISATION OF MICROBIAL CULTURES

WP 1 was focused on the degrader organisms, their physiology and their interactions with the matrix environment. Several contaminant degrading bacteria were characterised and new bac-teria were isolated including bacteria degrading phenoxy acid herbicides. Surprisingly certain waterworks sand filters were shown to have an inherent potential for degradation of selected pesticides including MCPA, bentazone, and BAM. The BAM degrading amidase gene of Amino-bacter sp. MSH1, bbdA was characterized in details and shown to have a very high affinity for BAM degradation (low KM value) explaining the high efficiency of this bacterium for BAM deg-radation at low concentrations. Attempts have also been done to immobilise degrading bacte-ria on specific carrier materials. Incorporation of cells, suspended in alginate, into pores of inorganic strong and rigid but porous material (intermediate carriers) were most efficient regarding 1) the highest degrading biomass immobilised, 2) the lowest bacterial release in column systems and 3) best performance in contaminant degradation tests. Based on this the intermediate carriers were selected to be further tested in WP5.

Selection and identification of microbial cultures
Metabolically degrading cultures. New methods for enrichment and isolation of bacteria min-eralising low contaminant concentrations were developed. The methods were tested with en-richments from a groundwater aquifer exposed to either high (25 mg/l) or low (0.1 mg/l) concentrations of the herbicide MCPA. Isolation of MCPA-degrading bacteria from the enrich-ments were then attempted by spreading cultures either directly on Petri dishes or on so called Low Flux Filters (LFF; 2.5 cm diameter, 0.22 μm pore size) overlaying the plates, thereby allow-ing a constant flux of low herbicide concentrations to the bacteria. Several MCPA degrading bacteria were isolated by the method and then further identified and characterised. Purity verification, catabolic gene contents and characteristics in regards to high/low MCPA concentrations was then undertaken.

Furthermore, enrichment cultures were established using filter material from different water-works sand filters. Sand filter material was assessed for the mineralisation of 14C-labeled MCPA, bentazone, and BAM. Mineralization of BAM was only found in sand filter material from Kluizen and Eeklo waterworks. Bentazone was mineralized in sand filter material from Kluizen, Eeklo, AWW, and Snellegem waterworks. In contrast, all tested samples, except samples from De Blankaart waterworks, showed mineralization of MCPA. We concluded that the tested samples show interesting pesticide mineralization activities. Especially, the occurrence of bentazone mineralization is of interest since pure cultures of bentazone mineralizing bacteria have not yet been reported.

Cometabolically degrading cultures. Several ammonia and methane oxidizing bacterial cultures were enriched and screened for their ability to degrade a selected group of micropollutants via cometabolic degradation pathways. Mainly sulfamethoxazole and benzotriazole were found to be degraded by methane oxidizing bacterial (MOB) enrichment cultures. The degradation, however, was always most rapid when no methane monooxygenase inhibitors or antibiotics were added, confirming the degradation being due to cometabolic activity by MOB having broad spectred methane oxidizing enzymes. Test for cometabolic degradation of micropollu-tants were also performed with several different commercially available ammonia oxidizing bacterial (AOB) cultures. The compounds tested were atenolol, BTZ, sulfamethoxazole (SMX), Chloro-BTZ (Cl-BTZ), mecoprop and diclofenac. These tests showed removal efficiencies for most of the compounds, but only partly due to cometabolic degradation. For BTZ a removal ef-ficiency of > 90% could be achieved in the biological treatment without inhibition of nitrifica-tion, but as the test with inhibitor present also showed a degradation of around 60%, it can be concluded that roughly 30% of the removal was due to cometabolic degradation by the present AOB. Some compounds such as atenolol showed heterotrophic degradation being the only rel-evant pathway for the disappearance of this compound, but SMX did only show removal when nitrification was taking place.

Based on the results the following bacteria and bacterial enrichment cultures where selected to be used for the development of the BIOTREAT remediation technologies:

• Aminobacter sp. MSH1 (BAM)
• Variovorax sp. SRS16 (Linuron)
• Sphingobium herbicidovorans MH (mecoprop)
• Novosphingobium sp. KN65.2 (carbofuran)
• Methane oxidising bacterial culture enriched from soil (MOBsoil )
• Methane oxidising bacterial enrichment culture (MOBbourgoyen)
• HANDS (ammonium oxidising enrichment culture)

Physiological and genetic characterisation of microbial cultures
The BAM degrading strain Aminobacter MSH1 was selected as a model bacterium in BIOTREAT and its physiological and genetic features have been characterized in details.

The genetic characterization and elucidation of the BAM degradation pathway in Aminobacter sp. MSH1 was initiated by identifying the amidase gene function performing the first step in BAM degradation. Moreover, a draft genome sequence of Aminobacter sp. MSH1 was ob-tained.

This information is important in elucidating the genetics of BAM degradation beyond 2,6-dichlorobenzoate and to elucidate functions involved in BAM degradation and its ability to cope with realistic low BAM concentrations at oligotrophic sand filter conditions. However, plasmid sequences were underrepresented in the draft genome and further sequencing (454 pyrose-quencing) of isolated plasmids of strain MSH1 was done. A draft sequence of the plasmid, pBAM2 was obtained by Illumina sequencing and annotation with RAST (Rapid Annotation using Subsystem Technology).

A BAM amidase gene (bbdA) was identified being responsible for the first step in the BAM deg-radation pathway in Aminobacter sp. MSH1. The BAM amidase has a size of 55.7 kDa and shows only limited homology to other known proteins. The most similar enzyme known is an enantioselective 2 phenylpropionamide amidase of Agrobacterium tumefaciens. The BAM degrading activity of bbdA was confirmed by cloning the bbdA gene in a pEXP5 CT/TOPO vector (Invitrogen) and expressing the protein in E. coli BL21(DE3)pLysS.

The Michaelis constant (KM) of bbdA was estimated by non-linear regression analysis using a Michaelis Menten equation as model and was calculated to be 0.69 µM. This value can be con-sidered as a measure for the affinity of the enzyme for its substrate BAM. This KM value is very low compared to other amidases, indicating that bbdA has a very high affinity for BAM. The high affinity explains why Aminobacter MSH1 degrades BAM so efficiently at very low environ-mentally relevant concentration.

At equimolar substrate levels the BAM amidase showed a 23.2 times higher conversion of OBAM and even a 50.3 times higher conversion of benzamide compared to BAM. These num-bers indicate that bbdA actually has a higher activity on benzamide and OBAM than on BAM, suggesting that this enzyme might actually be a benzamide amidase able to degrade BAM.

A method for quantifying the amidase gene using qPCR was developed and used in other WPs. Primers specific for the amidase gene were designed and tested on a dilution series of the gene and on samples with bbdA added to soil and sand filter DNA extracts - as low as 300 bbdA cop-ies per PCR reaction could be detected using the developed method.

In conclusion the biochemical characterization of the BAM degrading amidase of Aminobacter sp. MSH1, bbdA has been characterized in details. This enzyme has a very low KM value, indicat-ing a high affinity for BAM and interestingly also shows a higher activity towards benzamide and OBAM than towards BAM. Several proteins were upregulated in strain MSH1 in the pres-ence of BAM, many of which were also found to be absent in the dichlorobenzoic acid (DCBA) degradation impaired mutant M6.100g. This lead to the identification of 12 genes which we believe has a high chance of being involved in the BAM catabolic pathway. One of these genes was identified as bbdB, encoding the NADH dependent DCBA degradative protein bbdB.

Designing of carriers for cell immobilisation
The most simple bioaugmentation strategy is addition of degrader bacteria as a liquid culture to the environment. More advanced strategies using immobilised degrader bacteria may, however, be superior protecting the bacteria from predation by protozoa or preventing them from wash out from the environment. In the BIOTREAT project attempts have been done to immobilise degrader bacteria on different carrier materials to improve degradation efficiency of inoculated waterworks sand filters.

In general the carriers should not be toxic neither to humans or the entrapped bacteria. Several non-toxic carrier materials have been tested including alginate, chitosan and cellulose natural polymers and inorganic carriers as expanded clay, carbonates and nanosol gel matrixes based on silicone. A method has been developed to change the surface charge of bacteria facilitating efficient cell immobilisation. No significant change in the ratio between death and live Amino-bacter MSH1 or linuron degrading Variovorax SRS16 cells were observed following immobilisa-tion.

In general three principles for development of carriers can be determined: Bacteria can be at-tached on the surface of carriers, embedded in the matrix or they can be physically stabilized with specific techniques like freeze drying. According to the survivability and relatively easy up-scaling of the procedure, the embedding into the matrix is preferable. Similarly, stabilisation by modification of the bacterial surface by a Layer by layer (LBL) approach is according to our pre-vious experiments also a very good method for preserving the activity of cells.

Based on alginate matrix and LBL approaches different carriers have been prepared and were divided into three groups based on their complexity of the preparation: simple, intermediate, and advanced carriers.

The simple carriers were prepared exclusively based on the alginate matrix, which was modi-fied to increase strength, diffusion, specific density, and sorption capabilities. The best alginate beads were formed when extruding 2% alginate solution drop wise into 0.2 M CaCl2. The in-termediate carriers were prepared by incorporation of cells, suspended in alginate, into pores of inorganic strong and rigid but porous material. These carriers were prepared, because higher strength and durability may be needed to sustain higher pressures and abrasions in sand filters. Different porous mineral matrixes that have been used in aquarium filters were selected for development of such carriers. The advanced carriers were based on the LBL method for deposition of polymers. On such functionalized surface the bacterial cells were able to attach with strong electrostatic interactions.

Based on results from batch tests a selection of carriers were further tested for their suitability in a flow-through column system. The results from the column experiments are suggesting that the intermediate carrier (Seachem Matrix TM (MX)) is the best to be used in the waterworks sand filter. Also the simple carriers with quartz sand were effective for degradation of BAM. The same can be concluded from a financial perspective that preparation of the simple carriers is less expensive than the other procedures. Although the intermediate Sera Siporax (CV) carri-er was not tested in the column system, they were the best at the immobilizing the largest amount of bacteria, they had the lowest bacterial release and they performed best in batch mineralization tests. Thus intermediate CV carriers should also be considered as a prominent carrier.

Briefly, the procedure of preparing intermediate carriers is based on the suspension of cells dispersed in alginate solution of variable percentage (from 1% to 3%). The solution of alginate and bacteria is then forced to get incorporated into the pores of mineral matrix (e.g. MX) by the variation of pressure (from 0 – 20% of the normal atmospheric pressure). To stabilize the incorporated alginate-cell solution inside the mineral pores of mineral matrix the solution of CaCl2 is poured over the carriers or carriers are soaked into the solution.

WP2 PROCESSES – KINETICS OF DEGRADATION AND METABOLITE FORMATION

The overall aim of this part of BIOTREAT was to provide insight into the degradation processes fo-cusing on metabolic pathways and metabolite formation especially at the low micropollutant con-centrations found in groundwater and surface waters exploited as drinking water resources. A sensitive method for the analysis of assimilable organic carbon (AOC) at low concentrations was developed as this constituent may influence contaminant degradation efficiency. For the degrader bacteria studied it was demonstrated that parameters of substrate utilization kinetics estimated at high concentrations could accurately predict substrate utilization at lower concentrations under AOC-restricted conditions. Tentative degradation pathways have been established for several con-taminants including sulfamethoxazole (SMX), Oxcarbazepine (OXC) and carbofuran. Threshold concentrations below which degradation is either slow or not occurring may hamper degradation at low contaminant concentrations, but only for some bacteria as to example a high threshold val-ue was observed for carbofuran degradation by Novosphingobium sp. KN65.2 while such a threshold concentration could not be detected for BAM degradation by Aminobacter MSH1.

Microbial growth at low concentrations
Easily degradable organic carbon or as it is called assimilabe organic carbon is an important con-stituent that may control contaminant degradation. At one hand the AOC may serve as a growth substrate for the degrading bacteria thereby stimulating the degradation process on the other hand the AOC may be preferred as substrate at the expense of the contaminants. A method has been developed within BIOTREAT to estimate the AOC concentrations in water. The method has been applied to a system in which pure bacterial strains were growing on a target pesticide as the sole carbon source. In applying this method, growth on external sources of AOC was minimized and thereby enabling precise estimates of microbial growth and substrate utilization parameters for each strain-substrate pair. We first sought to validate the applied method by investigating growth and substrate utilization in high-concentration incubation experiments that paralleled con-ventional biodegradation assays. We then applied the method to a series of incubation experi-ments conducted at increasingly lower initial substrate and inoculum concentrations to determine the limits of microbial growth and substrate utilization. The bacteria studied included the previ-ously identified strains: Sphingobium herbicidovorans MH (mecoprop); Variovorax sp. SRS16 (linu-ron); Novosphingobium sp. KN65.2 (carbofuran); and Aminobacter sp. MSH1 (BAM).

The effects of increasingly lower substrate and inoculum concentrations on microbial growth and substrate utilization were unique for each strain. This is a fascinating and important result with respect to practical application of these strains in bioaugmentation processes; it is apparent that bacterial strains must be investigated on an individual basis and there is not a generalizable rule for bacterial growth at low concentrations of individual substrates. For strain SRS16, the initial substrate and inoculum concentrations had no effect on linuron degradation; all experiments re-sulted in removal of linuron to below the limit of detection within 14 days with a lag phase only apparent at the highest substrate (10 mg/l) and lowest inoculum concentrations (103 cells/ml). Growth of strain SRS16 was measurable in incubation experiments where the yield of cells was higher than the inoculum concentration. For strain KN65.2 both the initial substrate and cell con-centration had an apparent effect on the degradation of carbofuran. At cell inoculum concentra-tions of 103 and 104 cells/ml, no degradation of carbofuran was observed in incubations with initial substrate concentrations below 1 mg/l and 0.1 mg/l, respectively. At all other sets of initial condi-tions, degradation of carbofuran was complete to a concentration below the limit of detection within 28 days. Growth of strain KN65.2 was likewise measurable in those incubation experiments where the yield of cells was higher than the inoculum concentration. For strain MSH1, BAM disap-peared rapidly in incubations in which the inoculum concentration was high and the initial BAM concentration was low. However, as with the high-concentration incubation experiments, no cor-responding growth was observed while a persistent peak formed in the UV chromatogram puta-tively corresponding to 2,6-dichlorobenzoic acid suggesting a single, cometabolic biotransfor-mation reaction (data not shown). For strain MH, lowering the mecoprop concentration to below 10 mg/l nearly completely inhibited metabolic activity. Only at the highest initial substrate and inoculum concentrations was partial mecoprop utilization observed within 28 days.

Using these data, we estimated the kinetic parameters describing microbial growth and substrate utilization assuming Monod growth kinetics. The goal of the modelling exercise was to use a sim-ple model to estimate kinetic parameters for each strain-substrate pair that could subsequently be used to simulate biodegradation under varying sets of initial conditions. No kinetic parameters were estimated for strain MSH1 because no growth was observed. During the project we also pro-vided maximum specific substrate utilization rate ( ), the maximum growth rate ( ), and the sub-strate concentration giving one-half the maximum rate (K) for the remaining three strains. The estimated parameters were highly variable among the strains with varying over an order of magnitude and K varying over three orders of magnitude, indicating the unique kinetics of each growth process.

The next step was to determine whether or not microbial growth and substrate utilization kinetics estimated from high concentration incubation experiments could be used to predict strain behav-iour at environmentally relevant concentrations. We hypothesized that these observed shifts in kinetics are the result of a shift from single-substrate to mixed-substrate utilization kinetics as the concentration of the target substrate approaches the concentration of background assimilable or-ganic carbon (AOC). We reasoned that our method to measure substrate utilization and microbial growth under AOC-restricted conditions would eliminate mixed-substrate utilization and could re-solve these kinetic issues.

We first aimed to make robust estimates of Monod kinetic parameters for strains Variovorax sp. SRS16 utilizing linuron and Novosphingobium sp. KN65.2 utilizing carbofuran. The optimal dataset for kinetic parameter estimation was identified at an initial cell density of 105 cells ml 1 and initial substrate concentrations of 1 mg L 1 and 10 mg L 1 in experiments with strains SRS16 and KN65.2 respectively. Estimated parameter values and standard deviation of the marginal distributions are provided.

In general our results demonstrate that parameters of substrate utilization kinetics estimated at high concentrations can accurately predict substrate utilization at lower concentrations under AOC-restricted conditions. Multiphasic kinetics was not observed. While our data cannot disprove the existence of multiple, concentration-dependent uptake and transformation systems in bacteria that lead to observed shifts in substrate kinetics at low concentrations, our data is consistent with our hypothesis that observed shifts in kinetics could be the result of shifts from single substrate utilization to mixed substrate utilization. Published data suggest that shifts to mixed-substrate utilization can likewise result in shifts in kinetics.

Metabolite patterns at low substrate concentrations
The research about metabolite pattern at low substrate concentrations has primarily focused on 1) BAM degradation by Aminobacter MSH1, 2) carbofuran degradation by Novosphingobium strain KN65.2 3) sulfamethoxazole (SMX) degradation by a methan oxidizing bacteria (MOB) and 4) Ox-carbazepine (OXC) degradation in laboratory incubations with materials of waterworks sand filters.

BAM degradation: BAM is a highly stable metabolite from the herbicide dichlobenil and this resi-due is a well-known water contaminant in area where dichlobenil is or has been used. In order to determine mineralization of the compound at low concentrations, work has been done using radi-oactively labelled BAM. A Thin Layer Chromatography method with radiochemical detection was established, which made it possible to also look at the formation of BAM metabolites. An unknown metabolite was discovered and a relationship between the time needed for complete degradation of BAM and the extent of metabolite production was revealed. The degradation rate of BAM by Aminobacter sp. MSH1 can be controlled by adding different amounts of cells. There are examples showing slow BAM mineralisation, where the production of the unknown metabolite is below de-tection. Faster degradation could be obtained by using more MSH1 cells, but under these condi-tions the unknown polar metabolite is produced in higher levels, which could be a problem when bioaugmentation of drinking water resources is attended.

Attempts to identify the unknown metabolite failed, but the metabolite was shown to be degraded completely in waterworks sand filters as demonstrated in a survey including 6 different water-works sand filters. Based on these results it was concluded that the metabolite would not be a problem during bioaugmented remediation of BAM polluted groundwater.

Carbofuran degradation: In addition to BAM the degradation pathway of carbofuran by Novo-sphingobium strain KN65.2 was also elucidated. Several Novosphingobium mutants were produced and following incubation metabolite formation was characterized by high-resolution mass spec-trometry. Several transformation product structures were proposed and a single persistent me-tabolite was identified as 1-(2-benzoic- acid)-(1H,3H)-quinazoline-2,4-dione (BaQD).

Besides these metabolites it was also observed that several incubations turned reddish following prolonged incubations. One of the mutants was found to produce a more vibrant red colour in the incubations than the other or the wild type and from this incubation we were able to isolate sev-eral red metabolites. A chemical structure of two of the metabolites were suggested based on mass spectrometry and NMR spectroscopy

Sulfamethoxazole degradation: In addition metabolite formation from sulfamethoxazole (SMX) by a methan oxidizing bacterium (MOB) was studied in details. 4-nitrosulfa¬methoxazole (4-Nitroso-SMX) and 4-N-acetyl-SMX were detected, whereas in the experiments with heterotrophic bacteria only the formation of 4-N-acetyl-SMX was observed. Presumably these transformations do not significantly change the antibiotic efficiency as it is known that at least 4-N-acetyl-SMX has a simi-lar affinity as SMX to dihydropteroate synthase, the bacterial enzyme blocked by SMX.

Oxcarbazepine (OXC) degradation: OXC exhibited a half live of only 1.3 days in incubations with sand filter materials. Transformation products were isolated and enriched using semi-preparative HPLC/UV and measured via high-resolution mass spectrometry and NMR. Interestingly, the same transformation products were observed irrespectively of whether OXC or two carbamazepine pharmaceuticals (10OHCBZ and DiOHCBZ ) were used as target compounds.

Mechanisms of contaminant transformation at low substrate concentra-tions
The goal of this work was to describe mechanisms of transformation for model contaminants as related to the BIOTREAT project. Here, we present pathways and mechanisms based on plausible chemical and biochemical reactions. First, we examine the mineralization pathway for carbofuran. Carbofuran is utilized as a growth substrate for Novosphingobium sp. KN65.2 and several metabo-lites have been described (see above). Second, we examine the transformation pathway of ox-carbazepine and two metabolites of carbamazepine and oxcarbazepine by a mixed-community derived from a drinking water treatment plant sand filter. In presenting these two examples, we provide data and mechanisms for metabolic and co-metabolic processes.

Carbofuran: Based on identification of carbofuran degradation products, we proposed a tentative pathway to carbofuran mineralization by Novosphingobium sp. KN65.2:

The first step [1] in the proposed pathway is the hydrolysis of the carbamate to the corresponding alcohol, methylamine and carbon dioxide. As a second step [2], we suggest an electrophilic monooxygenation reaction analogous to ipso substitution reactions with para alkoxyphenols. Such a reaction can potentially be catalysed by flavin or cytochrome P450 dependent monooxygenases. Note that R-OOH represents the hydroperoxo-iron of a P450-dependent monooxygenase or the C4a-hydroperoxy-flavin intermediate of a flavin dependent monooxygenase. Step three [3] must be a reduction of the quinone (most likely NAD(P)H dependent) to the corresponding catechol analogous to reductions described for ortho quinones to hydroquinones. For further metabolism, we suggest meta cleavage [4] of the substituted catechol with subsequent hydrolysis [5] of the meta cleavage product to yield 3-hydroxy-3-methyl butanoic acid and 2-hydroxypenta-2,4-dienoic acid in analogy to a pathway described for the metabolism of 2-alkylphenols. Reaction [7] seems to be an unspecific hydroxylation reaction not necessarily involved in the productive pathway. Re-action [8] could be similar to reaction [7] or metabolite CAR199 could be a descendent of metabo-lite CAR238 carried through the pathway.

Oxcarbazepine: Interestingly, degradation of oxcarbazepine, 10OHCBZ and DiOHCBZ lead to the formation of the same TPs, even though differences in abundances of individual TPs were ob-served. This leads to the proposal of two main transformation pathways for DiOHCBZ, 10OHCBZ and OXC.

In general, transformation is initiated by dehydrogenation of DiOHCBZ, hydroxylation with subse-quent dehydrogenation for 10OHCBZ as well as a hydroxylation for OXC, giving rise to the for-mation of TP269A and/or TP269B (10,11-dihydoxy-carbamazepine and its hydroxyl ketone ana-logue). The formation of TP269A is succeeded by the cleavage of the C10-C11 bond leading to TP300. This dioxygenase-mediated dissociation has been also described in the biodegradation of catechol as well as phenanthrene. Subsequent ring closure via intramolecular reaction of the ni-trogen of the carbamoyl moiety with the carboxylic acid group and the loss of water leads to the formation of BaQD. Formation of TP269B is also followed by cleavage of the C10-C11 bond, lead-ing to the respective dialdehyde, and oxidation of one aldehyde to a carboxylic acid. Subsequent ring closure proceeds via intramolecular reaction of the primary amine with the remaining alde-hyde group which is followed by dehydrogenation (BaQM) and oxidation (BaQD).
Furthermore, the loss of the carbamoyl group of OXC, DiOHCBZ and 10OHCBZ was observed fol-lowed by oxidation (for DiOHCBZ) or hydroxylation followed by oxidation (for OXC and 10OHCBZ) at the C10 and/or C11 position leading to the formation of TP223B. From TP223B a benzilic acid rearrangement leads to a ring contraction. The consecutive elimination of the hydroxy group re-sults in the formation of 9-CA-ADIN. Decarboxylation, hydroxylation and oxidation of formed hy-droxyl-group to the respective ketone lead to the formation of ADON. As no further TPs were de-tected in the transformation experiments with ADON, mineralization and/or microbial uptake can be assumed for this TP. In addition, hydroxylation of one of the aromatic ring of 9-CA-ADIN gives rise to the formation of 4-OH-9CA-ADIN.

Threshold concentrations
Threshold concentrations below which degradation is either slow or not occurring were deter-mined for Sphingobium herbicidovorans MH, Novosphingobium sp. KN65.2 Variovorax sp. SRS16 and Aminobacter sp MSH1 degrading mecoprop, carbofuran, linuron and BAM.

Sphingobium herbicidovorans MH: In high-concentration incubation experiments, strain MH de-graded mecoprop rapidly to a residual concentration of 9.1±8.1 mg/l; the high standard deviation was the result of triplicate experiments yielding residual concentrations of 0.1 11.7 and 15.5 mg/l. In low-concentration incubation experiments, little or no mecoprop degradation was ob-served at initial concentrations below 10 mg/l. Based on the residual concentrations observed in the high-concentration incubation experiments, this apparent threshold concentration could have been expected. However, even when cells from the high-concentration incubation experiment yielding a residual concentration of 0.1 mg/l were inoculated into low-concentration incubation experiments, little or no mecoprop degradation was observed (data not shown). We suggest here that average residual concentrations observed in high-concentration experiments can serve as a valuable predictor of threshold concentrations in equivalent systems. This suggestion is predicated on the assumption that the fundamental mechanism resulting in both residual and threshold con-centrations is the same. We argue that this mechanism is biological in nature and related to an apparent dependence of the induction and expression of part of the metabolic pathway on the substrate concentration.

Variovorax sp. SRS16 and Sphingomonas sp. KN65.2: Strains SRS16 and KN65.2 at least partially degraded their respective substrates in the low-concentration incubation experiments. No thresh-old concentrations were evident for strain SRS16, while an apparent threshold concentration that was dependent on both the initial carbofuran and cell concentration was observed for strain KN65.2. In an effort to explain these observations, we used the kinetic parameters reported above to simulate the theoretical time course of substrate utilization for each strain at varying sets of initial substrate and inoculum concentrations.

We provide the simulated and measured data for strain KN65.2 utilizing 0.01 mg/l of carbofuran at varying initial cell concentrations. The simulated data shows that an inoculum of 106 cells/ml should theoretically degrade carbofuran completely after approximately seven days. Lower inocu-lum concentrations will theoretically result in residual carbofuran concentrations ranging between 20% and 100% after 28 days. On a purely conceptual level, the measured data align with this theo-retical expectation rather well, with complete utilization of carbofuran observed at an inoculum of 106 cells/ml and no utilization observed at inoculum concentrations of 104 and 103 cells/ml. We provided the simulated and measured data for 103 cells/ml of strain KN65.2 degrading carbofuran over a range of initial concentrations. The simulated data show that the rate of the process is strongly dependent on the initial carbofuran concentration within this range; carbofuran concen-trations ranging between 1 and 10 mg/l are predicted to degrade relatively rapidly while those in the range of 0.01 to 0.1 are predicted to degrade very slowly. Again, on a conceptual level the measured data align very well with the simulated data; very rapid degradation was observed in the range of 1 to 10 mg/l and no degradation was observed in the range of 0.01 to 0.03.

We also compared the simulated versus measured linuron utilization by strain SRS16 (results not shown). As with strain KN65.2 the measured data agreed very well with the theoretical simulated data on a conceptual level. Despite the disparate behaviour of strains SRS16 and KN65.2 at low initial substrate and inoculum concentrations, the experimental data could be explained by the unique substrate utilization kinetics estimated for each strain at high concentrations. This is an ex-citing and important observation in that growth and substrate utilization can be predicted at envi-ronmentally relevant concentrations with parameters estimated from relatively simple experi-ments conducted at substrate and cell concentrations that can be easily measured with common analytical techniques. Also, for strain KN65.2 we conclude that unique substrate utilization kinetics is the primary factor for the observed reduction of metabolic activity at low substrate concen-trations, and that a threshold concentration was not attained.

Aminobacter sp MSH1: It has been shown that BAM degradation is initiated by an amidase. This enzyme is constitutively expressed and a threshold concentration for induction hampering BAM degradation is therefore not a major issue.

WP3 SYSTEMS – M ICROBIAL INTERACTIONS WITH THE BIOFILTER ENVIRONMENT
This part of BIOTREAT examines the surrounding environment of the systems, focusing on the ef-fect of sand filter ecology on introduced degrader bacteria. Tools to analyse molecular fingerprint-ing results in order to provide ecological parameters including richness (Rr), dynamics (Dy) and functional organization (Fo) have been developed. It was concluded that BAM mineralization by MSH1 in a sand filter ecosystem can be affected by the presence of other sand filter bacteria but that this depends on the diversity. Even at relevant trace pollutant concentrations Aminobacter MSH1 was revealed to produce extracellular polymeric substances (EPS) which may necessary for biofilm formation. Furthermore, protozoa naturally living within waterworks sand filters was shown to predate on the MSH1 cells added the filters, thereby limiting BAM degradation. Copper (Cu) was demonstrated to control cometabolic degradation of several contaminants by methane-oxidizing cultures including chloropropham, metazachlor and benalaxyl. Cometabolic degradation of sulfamethoxazole (SMX) and carbamazepine (CBZ) by ammonium oxidizing enrichment cultures was also demonstrated.

Tools to describe the diversity of microbial communities in treatment filters
A microbial ecosystem, such as a waterworks sand filter consists of a microbial community inter-acting with its environment, including also the introduced contaminant degrading microbial spe-cies. The structure of microbial communities is commonly unravelled using molecular fingerprint-ing patterns and tools for interpretation such pattern, (e.g. PCR-DGGE) including the range-weighted richness (Rr), dynamics (Dy) and functional organization (Fo) parameters for the differ-ent experimental setups have been developed. Such interpretation provides an ecological and predictive value to the analysis of the structure and diversity of a microbial community in a given environment, enabling us to compare both different communities as well as the behaviour of a community facing different kinds of stress. An image analysis protocol optimised for the characterisation of biofilms composition and spatial structure and architecture (Ar) has been developed. Biofilm composition and architecture is typically analysed by use of cytostains combined with confocal scanning laser microscopy (CSLM). We have identified essential parameters to be extracted from the individual micrographs of the CSLM in order to describe spatial structure and architecture. The parameters includes biovolumes, substratum coverage, surface to volume ratios, thickness, roughness, relative population abundances, and population co-localization. Different image analysis software packages used to analyse CLSM micrographs were tested and ranked in relation to their ability to extract these parameters from individual micrographs. Among several programs, the Daime protocol was selected as the best-available package to be used. Daime is a scientific image analysis and visualization program for microbiology and microbial ecology offering tools for analyzing 2D and 3D microscopy datasets of microorganisms stained by FISH with rRNA-targeted probes or other fluorescence labelling techniques. Daime is a freely distributed stand-alone program, generated directly by the microbial ecology scientific community. It is the most comprehensive, compared to other tested packages, in terms of contained analysis features. It is relatively easy to use, and it combines image analysis with 3-D visualization functionality. Using the DAIME manual as a guideline, we have developed a user-friendly DAIME –based protocol to conduct image analysis on micrographs obtained with CSLM – combined with cytostains. The Daime user instructions were further modified to better describe the spatial structure of microbial biofilms of waterworks sand filters. Microbial community composition facing abiotic stress

Metabolical degradation processes Biofilm formation by the BAM-degrading Aminobacter MSH1 has been studied in details to better understand abiotic factors controlling surface colonisation of introduced bacteria. It has been shown that the bacterium produce extracellular polymeric sub-stances (EPS) even at relevant trace pollutant concentrations. EPS are high-molecular weight com-pounds secreted by microorganisms into their environment and known to be involved in biofilm formation. Various fluorescent stains have been tested for EPS labelling. By using the EPS stain ConA (Concanavalin A-specific) which binds to α-glucosepyranosyl residues and α- mannopyranosyl residues, at least part of the EPS was visualized in Aminobacter sp. MSH1 biofilm. A relatively larger volume of EPS was indeed visualized in biofilms fed with 1 µg/l and in biofilms fed with me-dia containing no other C source than in biofilms developed at higher BAM concentrations.

Then the question raised was whether BAM removal in biofilms fed at micropollutant concentra-tions was not due to abiotic loss. To examine this, biofilms were intervened with two spontaneous mutants of the BAM degrader Aminobacter sp. MSH1. The first mutant was M6 100g, a mutant that degrades only BAM and not DCBA; the second mutant was M1 100g, a mutant which can de-grade DCBA but not BAM. Both mutants were GFP labelled. By growing the MSH1-GFP that de-grades both BAM and DCBA and the two mutants in biofilms, we were able to investigate and to provide evidence about the role of sorption and other abiotic removal processes on BAM removal, both in the flow chambers and on the link of EPS with BAM presence/degradation. As expected, the strain MSH1-GFP degraded both BAM and DCBA, while the M6 100g-mutant degraded BAM but not DCBA. The M1 100g-mutant did not degrade BAM. BAM was clearly not adsorbed by the EPS. Clear differences existed in the number of cells of MSH1-GFP, M1 100g and M6 100g in the biofilms. MSH1-GFP, which metabolizes BAM and DCBA, had the largest number of cells. The BAM concentrations in the feed did not influence the number of cells in the effluent.

Confocal scanning images of the biofilms show that the biofilm structure depends not only on the BAM concentration but also on the composition of the medium. Even when the MSH1 was fed with artificial groundwater media, MSH1 formed a biofilm and degraded BAM. The amount of ex-tracellular polysaccharides though was less and was less loose, i.e. more closely associated with the colonies and attached to the glass surfaces. EPS was produced in biofilms without BAM indi-cating that they were not produced from the degradation of BAM.

Cometabolical degradation processes. The effect of copper (Cu) on cometabolical degradation of iopromide, bentazone, mecoprop, diclofenac, CBZ, and BTZ were studied in methane oxidizing cul-tures (MOB). Cu-limitation has previously been shown to stimulate co-metabolic degradation of organic chemicals by MOB. Significant cometabolic removal (81%) was only observed in Cu-starved cultures degrading BTZ. A linear correlation between the consumption of CH4 [mmol/l medium] versus the removed BTZ [mmol/l medium] was seen showing a clear link between BTZ degradation and CH4 oxidation activity of the culture.

Cometabolic degradation of metazachlor and benalaxyl by a MOB culture (Bourgoyen) expressing particle bound methane mono oxygenase (pMMO) was also demonstrated.

The microbial community of the MOBbourgoyen culture was identified using DNA extraction, followed by PCR and illumina sequencing and compared to an methane oxidizing culture enriched from soil (MOBsoil), The community structures of the investigated cultures were completely different. The main constituents of the MOBsoil culture originate from the Flavobacteriaceae, Chitinophagaceae, Bradyrhizobiaceae, Methylocystaceae, Methylophylaceae, Moraxellaceae and Xanthomonadaceae families while the main constituents of the MOBbourgoyen culture were Flavobacteriaceae, Chi-tinophagaceae, Methylophylaceae, Comamonadaceae and Methylococcaceae . For both cultures; all OTUs classified as methanotrophic Proteobacteria were classified into two MOB families: either Methylococcacae (Gammaproteobacteria or type I MOB) or Methylocystaceae (Alphaproteobacteria or type II MOB). However, while the Methylocystaceae family was an abundant community constituent of the MOBsoil (up to 8.8% of the total OTU count), it was not dominant in the MOBbourgoyen (0.07% of the total OTU count). On the other hand, the Methylococ-caceae family was a very important community constituent of the MOBbourgoyen (25% of the total OTU count) while it was less important in the soil culture (0.28% of the total OTU count).
The MOBsoil culture was enriched and used in the experiments with a copper concentration which allows sMMO expression and prevents pMMO expression (no added copper; Choi et al., 2003; Hanson et al., 1996) On the other hand, the MOBbourgoyen culture could oxidize methane at elevated Cu2+ concentrations of 10 µmol L-1 which is much higher than the sMMO inhibition concentration of 1 µmol L-1 reported by Begonja et al. (2001) and should eliminate any possible sMMO for-mation. Under these specific conditions, both cultures were able to co-metabolically degrade met-azachlor but at different pesticide degradation efficiencies. The sMMO expressing culture (MOBsoil culture) has a 7 times higher degradation efficiency than the pMMO expressing culture (MOBbour-goyen culture). It has been reported that the degradation rate of VC, t-DCE and TCE is much faster for sMMO expressing cells of M. trichosporium OB3b compared to pMMO expressing cells (Lee et al., 2006; Lontoh et al., 1998 and Oldenhuis et al., 1991; Han et al., 1999). Also considering the higher maximum inhibition concentration observed for sMMO expressing cultures (2.5 times high-er than pMMO), one might prefer the use of the sMMO expressing cultures over the use of pMMO expressing cultures. However, Lee et al. (2006) observed faster growth of pMMO expressing cells and faster degradation of VC, t-DCE and TCE when pollutant concentrations were higher than 100 µM. When pollutant concentrations were lower, sMMO was the preferred enzyme. Furthermore, it should not be forgotten that MOB only catalyse the oxidation step, which leads to the formation of unknown oxidation by products which can also have an effect on the maximum inhibition concentration (Han et al., 1999; Lontoh et al., 1999; Oldenhuis et al., 1991; Oldenhuis et al., 1989). The pMMO expressing MOB culture was also able to co-metabolically degrade chloropropham, benalaxyl and metazachlor. Until recently, oxidation of aromatic compounds has only been observed for sMMO. While pMMO is present in almost all methane oxidising bacteria, sMMO is mainly limited to type II MOBs. Both cultures contain representatives of type I as well as type II MOB, allowing these cultures to perform in a broader range of circumstances as type I and type II MOB show distinct ecophysiological features (Hanson et al., 1996) and have even been suggested to possess different life strategies (Ho et al., 2013). However, our illumina results show that type II MOB are much more dominant in the sMMO expressing soil MOB culture suggesting that sMMO is indeed expressed under the given limited Cu2+ concentrations. On the other hand, the pMMO expressing culture contains more type I MOB.

In conclusion, this study demonstrates for the first time the successful cometabolic degradation of chloropropham, metazachlor and benalaxyl by methane-oxidizing cultures. Both sMMO and pMMO expressing cultures seem to be able to degrade these components although the pesticide degradation efficiency for pMMO degrading cultures is lower at the tested pollutant concentra-tions. As MMO has a much lower substrate specificity compared to heterotrophic bacteria, the usage of methanotrophic bacteria to remediate contaminated water seems promising.

Cometabolic degradation of SMX and CBZ by ammonium oxidizing enrichment cultures deriving from either 1) sludge of a conventional municipal waste water treatment plant (WWTP), 2) acti-vated sludge of a waste water treatment plant of a hospital (Hospital WWTP), 3) soil (SOIL), or 4) water from the canal Coupure (COUPURE).

The culture originating from the soil sample showed up to 75% removal of SMX within 7 days, however, additional tests showed that a lot of the removal was also due to sorption (data not shown). The Hospital WWTP culture showed very good potential for degradation of CBZ with up to 64.6 ± 0.2% in 7 days.

Microbial community composition facing biotic stress

Degrader bacteria added to waterworks sand filters are faced with biotic stress in the form of competition with indigenous bacteria and grazing by protozoa. Much work in BIOTREAT aimed at applying the BAM-degrading bacterium Aminobacter sp. MSH1 to waterworks sand filters. We therefore study the protozoan grazing of MSH1 added to sand filter material.

The density of the bacteria added might have an effect on the protozoan numbers. In experiments added 107 MSH1 cells g-1 to filter sand, with an indigenous microbial community, resulted in a maximum of 5x104 protozoa ml-1. Adding 5x108 MSH1 cells g-1 resulted in a much higher density of protozoa implying that MSH1 was used as a food source for the protozoa in the sand filters.

A column experiment was also conducted in which MSH1 was inoculated in sand filter material that was either sterilized or containing the natural microbial community (i.e. bacteria and proto-zoa) of a sand filter. The results showed that BAM was removed more efficiently in the columns containing the sterile sand compared to the natural sand. Correspondingly, the number of Amino-bacter cells found in the sterilized sand was also higher, after the first 48 hours of the experiment, than the number of Aminobacter cells in the natural sand added MSH1.

In conclusion the results suggest that the protozoa naturally living within waterworks sand filters not only predate on the MSH1 cells added the filters, but also limit BAM degradation by MSH1.

Successful species invasion for bioaugmentation
The primary goal of this part of BIOTREAT was to point out the role of the structure of the whole microbial community in counteracting the effect of a new degrader species invasion and how this can influence specific ecosystem functionality. Still the basic process occurring during invasion of an existing community are not yet fully understood. To investigate the role of the evenness on in-vasion, artificially prepared communities with a constant diversity but varying evenness were used together with gfp-labelled model-invaders. The more uneven a community was the higher was the degree of invasion. So for bioaugmentation, i.e. the desired invasion of species, a lower degree of evenness of the receiving community would seem favourable. However, in the tested system, in-vasion also decreased the indigenous functionality as such, which in a lot of cases could be prob-lematic for implementation. Similar tests but this time using sand filter isolates supported the ear-lier findings that invasion was in general more successful in less even communities, but the degree of invasion also showed a high dependency to the actual dominant species. Most probably this is due to the level of direct competition between the invader and the dominant species. This needs to be kept in mind when considering a certain species for bioaugmentation.

Biofilm experiments were also performed to examine whether the BAM degrading strain Amino-bacter sp. MSH1 was able to invade a sand filter microbial biofilm community. To study this, bio-films of the gfp-labelled variant of MSH1 in the presence of a microbial community extracted from an operational sand filter in Sinaai (Belgium) were developed in flow chambers continuously fed with a medium with different concentrations of BAM. At all tested conditions a MSH1 mono-species biofilms developed and biomass of MSH1 decreased with decreasing BAM concentrations. In the presence of the sand filter community, the numbers of MSH1 cells appear lower compared to that found in mono-species MSH1 biofilms. In the flow chambers inoculated with MSH1 and the sand filter community which was continuously fed with 1 mg L-1 BAM, MSH1 formed small isolated colonies within the multi-species biofilm of sand filter bacteria. This contrasted with the corre-sponding mono-species biofilms of MSH1 in which the MSH1 cells spread out over the entire sur-face. MSH1 cells appear to avoid contact with the sand filter bacteria.

No BAM degradation was observed in the flow chambers containing only sand filter bacteria. This coincided with the inability of the Sinaai sand filter community to mineralize BAM. BAM degrada-tion was observed in flow chambers inoculated with only MSH1 and in flow chambers inoculated with both MSH1 and the sand filter community. The extent of BAM degradation at steady-state conditions was however on average 10-20% lower in the mixed biofilms compared to the mono-species MSH1 biofilms indicating that either the BAM degrading activity and/or the growth of MSH1 was affected in the mixed species biofilm systems.
We conclude that we have first indications that BAM mineralization by MSH1 in a sand filter eco-system can be affected by the presence of other sand filter bacteria but that this depends on the diversity. A higher richness of sand filter bacteria seems to wipe out effects observed in the pres-ence of single strains of sand filter bacteria. We can also conclude that MSH1 is able to establish in biofilms of sand filter communities and perform its BAM degrading functionality even at 1 μg/l influent concentrations although the presence of the community affects degradation negatively.

WP4 MODELS AND PREDICTABILITY – MODELLING AND PREDICTION OF MICROBIAL DEGRADATION OF MICROPOLLUTANTS
Mathematical modelling of biodegradation kinetics can improve our understanding of the reac-tions and allows the design of in situ or in reactor applications of microbial remediation. In this part of BIOTREAT we have chosen, validated and parameterised appropriate bio-kinetics expres-sions for 1) metabolic primary pollutant removal of trace pollutant concentrations and 2) cometa-bolic oxidation of trace pollutant concentrations by ammonium oxidizing bacteria (AOB).

Choose, validate, and parameterize appropriate biokinetic expressions for cometabolic oxidation of micropollutants by AOB and MOB

In this part of BIOTREAT a generalized model review for cometabolic transformation of micropollutants was conducted. We successfully generated the first set of cometabolic models based on the metabolic relationship of non-growth and growth substrates. We chose three criteria to evaluate the cometabolic models for the considered use: ability to fit experimental data, identifiability of parameters and suitability of the model size and complexity. The ability to fit to experimental data and the identifiability of the model parameters are of high importance for any model to be valuable. To make decisions about the most suitable models for the future, we evaluated the following five models for their ability to fit the synthetic experimental observations and whether they would return reasonable parameter estimates: A 1st-order model, a Monod model, a reductant model, a competition model and a combined mode (see deliverable 4.1 for details).
Simulated batch experiments were designed to resemble actual cometabolic substrate degradation experiments. These observation were generated by using the most complex model structure and assumed parameters based on the literature reported range. The weighted sum of squared residuals (WRSS) of the reductant model in the parameter/initial concentration domain is consistently lower than in the other three models. WRSS are used to find parameter values that minimise a measure of badness of fit originating from the fact that model equations are generally nonlinear forming a non-linear optimization problem. Akaike information criterion (AIC) and Bayesian information criterion (BIC) results show that the first-order model was smaller than the other models because it has less parameters. Besides the first-order model, the reductant model was much better than the other models for the AIC and BIC criteria. AIC and BIC are criteria used to rank models in accordance to complexity and size. The identifiability of the reductant model was better in the higher initial cometabolic substrate concentration zones, while in the lower ini-tial cometabolic substrate concentration zone, the first-order model was much better than the other models. From our simulations and model comparison, we can choose the best suitable mod-el for different parameter/initial substrate concentration condition domain. While the first-order model is at times superior (for it has fewer parameter than other models), it yields poor weighted sum of squared residuals. We conclude that the Reductant model is most adequate to describe observations across the considered experimental domain.

Choose, validate, and parameterize appropriate biokinetic expressions for primary pollutant removal at ‘micro g/l’ range
The concentration of easily assimilable organic carbon (AOC) largely determines the microbiologi-cal stability of drinking water (Egli 2010). Residual concentrations can be caused by diffusion limi-tations, maintenance energy demand of the organisms, a threshold concentration for enzyme in-duction, and environmental factors, like the presence of additional substrates. With respect to pesticide removal strains, many previous studies focused on their survival and/or die-off with pes-ticide as the sole carbon source, whereas only little knowledge is available on factors affecting their growth under environmental conditions, i.e. growth at low cell/nutrient concentrations with mixtures of substrates, e.g. groundwater (~10 µg/l AOC), shallow stream (~100 µg/l AOC), and stagnant pond water (~1000 µg/l AOC). In this research part, we developed a comprehensive gen-eral dynamic model framework including growth/survival state of the specific pesticide degrading strains in the environment with the target trace pollutant, an aquatic background AOC and back-ground bacterial community. The model describes different reactions and different relationships between the specific pesticide degrading bacterium and the background bacteria under different background AOC that could support growth. We carried out the model framework to explain the competition between special strain and the background community; the competition between the AOC and the transformation products (TP) and also the relationship between the special strain with the AOC and TP, which were difficult explaining experimentally.

Incorporate Biokinetic Rate Expressions in Continuum Biofilm Models to infer Reactor Scale Models for Cometabolic Micropollutant Oxidation

A continuum biofilm-bioreactor model that describes nitrification in a pilot-scale rapid sand filter has been developed (Deliverable D4.2). This model has been enhanced by explicit inclusion of ap-propriate biokinetic terms to capture co–metabolism.

The relevant additional process is described in the third row, entitled ‘cometabolic degradation’. In this process cometabolic substrate, Sc, is removed at the expense of ammonium oxidizing biomass, XB, AOB and this dependency is characterized by a transformation yield. ‘T. T’ quantifies the units of biomass consumed per unit of cometabolic substrate removed. It is further, assumed that Monod kinetics are applicable to describe the removal kinetics.

The model was encoded in the free open source software Aquasim (www.aquasim.eawag.ch).
For model implementation, specific operation and design parameters for the pilot-scale rapid sand filtration were taken from Albers et al. (2015). The pilot column had a diameter of 30 cm and a depth of 80 cm sitting on top of 30 cm of support material. The filters were operated at an average hydraulic loading rate of 3.7 m h-1. Based on our earlier experience, the pilot scale column was modelled as three completely mixed biofilm reactor compartments in series. Estimates of biofilm area, biofilm thickness, and other important system biokinetics and hydrodynamic parameters were based on available knowledge or best estimates. All details are provided in Deliverable 4.4.

Incorporate a Model for Oligotrophic Growth on micropollutant with a Biofilm Dynamics model in a multipopulation reactor-scale process model

The above mentioned continuum biofilm-bioreactor model that describes nitrification in a pilot-scale rapid sand filter (deliverable D4.2) were also enhanced by explicit consideration of an addi-tional process: an oligotrophic strain that specifically degrades a trace pollutant.

The relevant additional process is described in the third row, entitled ‘growth of ‘XB, MSH1’. In this process the trace pollutant substrate, SBAM, is removed by the specific degrader XB, MSH1, and this removal results in growth of XB, MSH1 captured by a growth yield YMSH1. YMSH1 quantifies the units of biomass produced per unit of substrate processed. It is further, assumed that Monod kinetics are applicable to describe the removal kinetics. In addition to growth, the strain is subject to loss (last row in Matrix): this term includes all possible loss mechanisms such as loss via endogenous decay, biomass detachment, activity loss, protozoan grazing, etc.

This model was encoded in the free open source software Aquasim (www.aquasim.eawag.ch) and the model implementation, specific operation and design parameters for the pilot-scale rapid sand filtration were taken from Albers et al. (2015) as described above. In Albers et al. (2015), the BAM degrading Aminobacter MSH1, was inoculated in the columns to stimulate BAM degradation via bioaugmentation. This strain was initially mixed in the top 20 cm of the column; hence the top 20 cm was modelled as two equally sized completely mixed biofilm reactor compartments of 10 cm depth, followed by a third reactor of 60 cm depth. Estimates of biofilm area, biofilm thickness, and other important system biokinetics and hydrodynamic parameters were again based on available knowledge or best estimates.

All details are provided in Deliverable 4.3 and Deliverable 4.4.

WP5 APPLICATION AND RELIABILITY – EVALUATION AND ADJUSTMENT OF BIOTREAT SYSTEMS FOR LARGE SCALE APPLICATIONS
The focus of this part of BIOTREAT was on large-scale application of the developed technologies. The research included laboratory-scale reactors and column experiments carried out in the laboratory and at waterworks. The metabolic remediation strategy used the BAM-degrading Aminobacter sp. MSH1 as model organism for remediation of BAM-polluted water, while the cometabolic strategy used the ammonium oxidizing bacterial (AOB)-mixed culture HANDS to remediate water polluted by carbamazepine (CBZ), sulfamethoxazole (SMX) and benzotriazole (BTZ). Inoculation with the BAM-degrading Aminobacter sp. MSH1 resulted in up to 75% removal of 0.2 µg/l BAM giving concentrations in the purified water well below the 0.1 µg/l legal threshold limit. Though, it was difficult to maintain efficient degradation for longer time periods due to loss of MSH1-bacteria. Loss of bacteria was diminished by immobilising the degrader bacteria on the ‘intermediate carriers developed within WP1. No clear decrease in micropollutant concentrations was observed at the large-scale cometabolic facility added the HANDS culture. This was attributed to the short contact time (20 – 30 min) and the relatively low N/micropollutant ratio of waterworks sand filters.

Larger scale immobilised systems for treating pesticide-contaminated groundwater at waterworks for drinking water production by means of metabolic processes
In this part of BIOTREAT several pilot scale sand filter systems have been in operation to test the efficiency of the BIOTREAT metabolic strategy for remediation of pesticide polluted drinking wa-ter. The system established in Denmark consists of an aeration basin followed by two rapid sand filters that could be operated in parallel or series. Ports for water sampling were placed before aeration for raw water analysis; between the aeration basin and filter 1 (inlet water), after filter 1 (filter 1 outlet), and after filter 2 (filter 2 outlet). Backwashing could be performed at desired time intervals.

The pilot waterworks was established 15 km west of Copenhagen and received anaerobic ground water from two waterworks abstraction wells, both polluted by BAM (0.13-0.22 µg/l BAM).

In the first experimental phase, quartz sand from the remediation plant in operation with a natural population of microorganisms (filter 1) was compared to fresh quartz sand (filter 2). Inoculation with MSH1 resulted in an immediate drop in BAM concentrations from 0.15 to 0.08 µg/l in both filters corresponding to a reduction of 50% during the first two days of operation. With time, how-ever, the capacity to degrade BAM decreased and after 23 days only 5-11% of the inlet BAM con-centration was degraded, most in filter 1. At the same time that the capacity for BAM degradation disappeared, the numbers of MSH1-bacteria in the filters also decreased.

In the second experimental phase, removal of BAM in a filter with the high porosity products Fil-tralite NC and Filtralite MC (filter 1) was compared to filter 2 with fresh quarts sand similar to the sand used in filter 2 in phase 1. As observed in phase 1, inoculation with MSH1 resulted in an im-mediate reduction in the outlet BAM concentration to well below the legal threshold limit. The, BAM removal was most efficient in the Filtralite filter, to which also sorption occurred. In both fil-ters the capacity to remove BAM declined with time, most rapidly in the filter with the Filtralite products. As in the first phase this decline coincided by a significant reduction in MSH1-bacteria, especially from the top part of the filters were also the highest densities of bacteria were seen. Predation on MSH1 cells was indicated by a twofold increase in the protozoan density.

In phase 3, backwashing was avoided in filter 2 by changing from parallel to serial operation mode, where precipitated iron oxides were removed by backwashing in filter 1 only. This was done to investigate whether loss of MSH1-bacteria could be diminished ensuring longer maintenance of the capacity for BAM degradation. Following inoculation, the BAM concentration decreased to below 0.1 µg/l in the outlet water of filter 1 corresponding to a reduction of about 50%. A further 50% reduction was achieved in filter 2 giving overall removal efficiency in the system of 75%. With time, however, the BAM-removal capacity decreased, though most rapidly in the backwashed fil-ter 1. Also the number of protozoa increased after MSH1 inoculation. In filter 1 (filter material not exchanged between phases 2 and 3) the response was fastest with most of the increase occuring within a day after inoculation in both top and depth. Hereafter the number of protozoa decreased along with the decrease in MSH1-bacteria.

In conclusion inoculation with the BAM-degrading Aminobacter sp. MSH1 resulted in up to 75% removal of 0.2 µg/l BAM giving concentrations in the purified water well below the 0.1 µg/l legal threshold limit. In addition no unwanted BAM degradation products were observed and no effects of inoculation on important filter processes like oxidation of iron and ammonium were seen. Though, it was difficult to maintain efficient degradation for longer time periods due to loss of MSH1-bacteria. Significant losses of bacteria were especially observed during backwashing. By avoiding backwash procedures the degradation was prolonged but bacteria and hence degradation activity was still lost with time.

A pilot scale system was also established at a drinking water production plant of ‘De Watergroep’ in Egenhoven, Belgium. Similar to the Danish facility, the bioremediation strategy was based on bioaugmentation with the BAM degrading Aminobacter sp. MSH1. In this facility it was investigat-ed whether bacteria entrapped in alginate and immobilized on a porous carrier (Seachem Matrix; input from WP1; see chapter 3.1.3) improved the BAM degrading efficiency.

Before inoculation, the MSH1 cell suspension and MSH1 containing carriers were tested for BAM mineralizing activity. BAM mineralization was much faster for the MSH1 carriers as after one day already 20% of the added 14C-BAM was mineralized as opposed to 10% for the free suspended cells. However, the BAM mineralization percentage increased to a greater extent the four follow-ing days for the free suspended cells compared to the carriers.

BAM in the influent and the effluent of the four columns of the pilot scale system was regularly measured. In the first pilot scale experiment, the column 2, bioaugmented with free suspended MSH1 cells, showed clear degradation of BAM during the first two weeks after bioaugmentation, reaching in the first week even 84.8% degradation. However, starting from day 17 after bioaug-mentation, the difference between the BAM concentration in the influent and the effluent became gradually smaller. In the column 3, containing the MSH1 containing carriers, residual BAM concentrations in the effluent were lower compared to the influent indicating also degradation of BAM. Moreover, during the 44 operational days, the BAM concentration in the effluent was al-ways below the norm of 0.1 µg/l . As time progressed, an increasing trend in BAM concentration in the effluent of column 3 with regard to the influent concentration was observed. However, on day 44 there was still 52.3% degradation in column 3 as opposed to only 18.7% degradation in col-umn 2. BAM analysis for control columns 1 and 4 resulted in no statistically significant difference from the results of the influent concentration. It is therefore concluded there was no BAM degra-dation in these columns.
In the second pilot experiment, column 3 showed the same BAM removal as observed in the first experiment. Interesting is that the MSH1 carriers were already 6 months old (stored at 4°C) and apparently did not diminish in BAM-degrading activity as the freshly made. Column 2 showed im-proved BAM removal, i.e. 50% compared to the first experiment (30%), which could be related to an improved BAM mineralization capacity and/or the recirculation.
The effluent of the two inoculated columns (column 2 and 3) was regularly sampled and loss of Aminobacter cells from the filters was determined by quantifying the numbers of the BAM cata-bolic genes bbdA and bbdB by qPCR in the effluent.
Samples of the sand filter material at the top (20 cm) of the columns were regularly taken after bioaugmentation. For column 3, these samples contained some MSH1 carriers; however, at this point DNA was only extracted from sand and not from carriers. The sand filter samples were used to quantify the presence of MSH1, bbdA and bbdB over time with qPCR.

To conclude, it was observed that the presence of MSH1 in the top of the columns decreased sig-nificantly during the first two weeks of operation but to a lesser extent for column 3. The MSH1 cells in column 3 have a higher BAM mineralization capacity than those in column 2, since bbdB was proportionally more present.

It can be concluded that the bioaugmentation with MSH1 in the sand filter columns by using MSH1 immobilized carriers showed most promising results. Degradation of BAM to concentrations below the norm were achieved for at least a period of 4-5 weeks in both pilot scale experiments for col-umn 3 at an average of 70%. In column 2, to which free MSH1 cells were added, BAM was only degraded efficiently in the first two weeks in the first experiment with an average BAM removal of 30% in the first 30 days of operation. In the second pilot scale experiment, BAM removal was improved to 50% by recirculating the inoculum for 5 days over the column before starting normal operations. The inoculation of free suspended cells in the sand filters is too inefficient at this point as most cells are immediately lost due to shear and bad initial attachment of MSH1 cells to sand. Carriers provided considerable improvement for reducing the loss of cells during inoculation. However, design of the carriers and the pilot system should be further improved to guarantee a long term (6 – 12 months) operation.

Immobilised systems for treating water contaminated with pharmaceu-ticals by means of cometabolic processes
Medium-scale sand filter reactors were also operated to investigate cometabolic degradation of micropollutants by ammonium-oxidizing bacteria (AOB). One sand filter acted as a control reactor, other sand filters were test reactors in which the co-metabolic conversion of micropollutants by means of AOB’s was tested under different operation conditions. Both reactors were always fed with artificial water, i.e. local tap-water with addition of micropollutants and, for the test reactor, with addition of 2 to 4 mg ammonium-N/l.

At the start, the test sand filters were seeded with an AOB-mixed culture designated HANDS. The culture was previously tested in batch lab-scale experiments and showed the capacity for co-metabolic conversion of certain micropollutants. Three micropollutants were selected and added to the influent of the sand filters, namely carbamazepine (CBZ), sulfamethoxazole (SMX) and ben-zotriazole (BTZ). SMX and BTZ were previously shown to be co-oxidized by AOBs while no conver-sion was obtained with CBZ (acting as a negative control in the sand filter tests).

No problems were ever encountered with the nitrification in the test sand filter: the latter re-mained stable and complete under the different process conditions tested (continuous feeding with artificial water, containing 6 to 8 mg O2/l). The nitrification was not only demonstrated by means of the frequent N analyses (> 95% NH4+ conversion), but also by means of a clear drop in dissolved oxygen (DO) and in pH (compared to the control reactor).

The operation parameters of the two sand filters that were altered in the course of the experi-ment were the influent flow rates and hence the hydraulic retention times (HRT), and the mi-cropollutant concentrations. Since it considered rapid sand filters, a short HRT of 20 to 40 min was aimed at. The dosages of the micropollutants were started at relatively high concentrations (100 µg/l) as also used in the lab-scale batch tests. During the test, the micropollutant concentrations in the influents were stepwise lowered down to 0.3 µg/l (of each micropollutant). Since the ammoni-um-N concentration in the influent of the test sand filter could not be increased because of the limited amount of oxygen in the influent water (in the range of 8 mg O2/l), a change in influent flow rate was always accompanied by a change in N-load.

Although it could be demonstrated in several batch tests that the biomass in the test pilot sand filter reactor still had the capacity for co-metabolic conversion of the micropollutants tested, no clear decrease of these could ever be measured in the continuous sand filters. Hence, it had to be concluded that the specific process conditions in these sand filter reactors did not allow for the co-metabolic conversion of the micropollutants investigated. Limiting factors for this co-metabolic conversion in a continuous sand filter reactor are most probably the short contact time (20 – 30 min) and the relatively low N/micropollutant ratio. Further batch tests confirmed that the colo-nized sand still had co-metabolic activity for the removal of SMX. Depending on the influent N-concentration and, consequently, on the N-load and N/micropollutant ratio, the co-metabolic con-version was relatively low and ranged between 6% and 22%. A comparison between the conditions in the batch tests and the applied operational parameters in the pilot scale test showed that, even if the very long contact times of 4 days would be feasible, the theoretical co-metabolic removal of the investigated micropollutants would still be too low to give removal efficiencies which would justify any effort to apply this technology in practice. Therefore, it was concluded that, although the biomass in the test sand filter still contained co-metabolic activity, the overall process conditions of a sand filter did not allow for co-metabolic conversion of the micropollutants tested.

WP6 INTEGRATED ASSESSMENT AND PERFORMANCE VALIDATION
With the view to developing a water purification technology that is also competitive it is necessary to analyse both the financial and the environmental aspects of its further exploitation. In BIO-TREAT financial and environmental aspects of the developed technologies have been assessed us-ing cost-benefit analysis (CBA) and life cycle impact assessment (LCIA) approaches. It is concluded that well relocation is the cheapest option also having the lowest environmental impact. This op-tion, however, depends on the distance to unpolluted water resources and is often not possible. The CBA showed that the BIOTREAT metabolic technology, especially used in combination with the BIOTREAT carrier is the most attractive technology especially for small scale drinking water pro-duction plants due to the relatively low investment costs. The BIOTREAT metabolic technology had an environmental impact similar to granular activated carbon GAC treatment. Immobilisation of degrading bacteria on carriers increased the environmental impact, due to the additional materials needed, but probably the impact could be reduced using other carrier materials as to example ordinary sand.

Performance and cost analysis
The final goal of the BIOTREAT project was the development of a prototype system ready for commercialisation. The cost-benefit analysis (CBA) was performed to check whether commerciali-sation is a viable option.

The first step in the CBA was the collection of data including inputs from the small-scale tests car-ried out mainly in WP3 and the large scale prototype test carried out mainly in WP5 (see chapter 3.3 and 3.5). Financial data were gathered from the drinking water production experts mainly from the drinking water companies involved in the project. The next step was to build in the col-lected data in a CBA model. The model was then used to calculate the costs for the following four scenarios:

• BIOTREAT bioaugmentation technology: in this scenario a Danish waterworks is upgraded with the bio-augmentation technology under development in this project. In particular, sand filters are inoculated with BAM-degrading bacteria. After the sand filter, a UV disinfection step is added, to make sure the number of bacteria leaving the filters does not exceed the legal threshold. Every other aspect of the waterworks operation remains unchanged.
• BIOTREAT bioaugmentation technology + carrier: This scenario is essentially equivalent to the one above, with the difference that carrier materials are used to improve the stability of the bacterial population. The materials used are a mineral substrate and sodium alginate.
• Granular activated carbon (GAC) filter: GAC has been identified as the most likely competitor technology for BIOTREAT. In fact, several waterworks in Denmark already apply this technolo-gy in order to remove micropollutants. In this scenario, a GAC filter bed is installed and operat-ed in an existing Danish waterworks, followed by a UV disinfection system, to ensure microbio-logical safety. Every other aspect of the waterworks operation remains unchanged.
• Well relocation: a possibility for a waterworks that risks exceeding the legal limits for mi-cropollutants is to close the contaminated wells and open new ones elsewhere, with concen-trations of micropollutants under the legal limits. In this scenario, the waterworks operation is not changed with any additional process, but it involves dismantling the existing wells and opening new ones.

Part of the data interpretation was the sensitivity analysis by which we could identify the factors that influence the costs the most. Giving answer to the questions:

• How do the four scenarios relate to each other financially?
• Is the BIOTREAT technology financially competitive with alternative approaches?

The results of the CBA, calculated for a drinking water production plant producing 800,000 m3 per year, are given as investment costs/yearly costs, as well as yearly operating costs.

Based on the sensitivity analysis it is concluded that the calculated values for the BIOTREAT bio-augmentation technology scenario and BIOTREAT bio-augmentation technology + carrier scenario cannot be given within the 30% deviation range. Further research is needed to get more reliable data on the amount of inoculum needed per m3 drinking water, the costs of one litre high-density BAM-degrading bacteria and the (regeneration) costs of the carrier.

Based on the available data it is concluded that the well relocation scenario is the cheapest scenario. But because the length of the transportation pipe has a large effect on the costs of the well replacing scenario, the well replacement scenario is only cheaper if the new well is placed within a few kilometres from the water production plant (3.5 kilometres for the 800,000 m3 per year scenario). If clean water is more than a few kilometres away, implementing a BAM removal strategy becomes the best solution.

For the larger drinking water production volumes the GAC filter is the cheapest BAM removal technique. The BIOTREAT technology, especially the BIOTREAT bio-augmentation technology + carrier is the most attractive technology for small scale drinking water production plants due to the relatively low investment costs.

Because the data for the CBA is not reliable enough, it cannot be concluded that the BIOTREAT technology is ready for commercialisation. However, the technology looks promising from a finan-cial point of view. If further commercialisation is attempted, it is advised to focus on relative small drinking water production plants where clean water is more than a few kilometres away.

In general, the amount and quality of primary data available for the CBA was insufficient to deliver an accurate picture of the BIOTREAT technology. This is not the fault of the project, since lack of data and uncertainty is typically inherent to any CBA applied to an emerging technology. There is no experience of applying this technology in waterworks, and the field tests carried out were limited. As a consequence, many assumptions had to be made in the study, and more often than not, expert judgement rather than recorded data had to be used. This is the reason why in the conclusions summarised in the previous sections we generally abstain from providing quantitative assertions, but instead give qualitative judgements.

We think nevertheless that this study has shed light on the potential environmental impact and the market potential of the BIOTREAT technology, benchmarking it against current alternatives and pointing out key areas where either more knowledge is needed or attention should be put in order to minimize the environmental impacts and costs of this technology.

Elaboration of Life Cycle Assessment (LCIA)
The life cycle impact assessment (LCIA) for the four scenarios (see chapter 3.6.1) under study, ac-cording to 15 impact categories has been studied. The basis for comparison, or functional unit, is the provision of 1 m3 of drinking water with BAM levels under the legal threshold of 0.1 μg/l. The graph shows the LCIA scores in relative numbers, whereby for each impact category the highest scoring alternative is normalized to a score of 1, and the remaining ones, are expressed in relation to that score. In terms of interpreting the graph, the higher the LCIA score the higher the environ-mental impact.

The main aspects to highlight are the following:

• The BIOTREAT-carrier alternative appears as the one with the highest environmental impact, in all categories but the one on aquatic eutrophication, where BIOTREAT has a higher environ-mental impact. The relatively high impact of BIOTREAT-carrier is associated with production of the mineral carrier material.
• The BIOTREAT alternative obtains somewhat lower impact scores than the GAC alternative, with the exception of the impact categories of land occupation and aquatic eutrophication. The higher impact in these categories is mainly related to the production of the bacterial inoculum.
• In all 15 impact categories, the best performance is attributed to the well-relocation alterna-tive. This can be explained by the fact that even though shutting down a well and opening a new one is material- and energy-intensive, it is a single operation where the lifespan of the new well is 50 years, and as opposed to the other alternatives, it does not involve any addi-tional and continuous input of energy or chemicals to treat polluted water.

At this level, i.e. midpoint, where all impact categories are assessed separately, we can identify well re-location as the best alternative from the ones assessed. From the remaining alternatives, which involve tackling water pollution rather than just avoiding it, it is interesting to see that BIO-TREAT is approximately at the same impact level as GAC, as long as carriers are not used. We delve into the carriers issue in next the section.

Sensitivity analysis on carrier materials

As we have seen, in most impact categories the use of carrier materials leads to the BIOTREAT-carrier alternative being the least desirable. It must be taken into account that assessing this alter-native leads to uncertain results, given that:
• At this early stage of technology development it is not clear what type of mineral carrier would be used in a full-scale application.
• The amount of mineral carrier, alginate, and CaCl2 needed per m3 of treated water is based on laboratory tests, which are far from being optimized when compared to a full-scale ap-plication.

We have checked the influence in the results of considering a different carrier material. Based on conversations with Institut za Mikrobioloske Znanostiin Techologije who have developed the carri-ers, it was decided that the most environmentally friendly option would be to use sand (instead of expanded clay) extracted from the waterworks site. In this way, we use a less energy-intensive material and avoid the transport step.

The results of this sensitivity analysis are shown for greenhouse-gas emissions (GHG) (see figure 39 in deliverable 8.16). In this figure the first bar shows the overall emissions for BIOTREAT-carrier, with our initial assumption on expanded clay, whereas the next one shows the emissions when local sand is considered instead. The remaining bars show the emissions for the other alternatives, for reference. It can be seen that the choice of carrier material has a substantial effect on the GHG-intensity of BIOTREAT-carrier. When local sand is considered, the GHG emissions are reduced by more than 50%. At this point, BIOTREAT-carrier is approximately at the same level of impact than BIOTREAT and GAC.

Finally, a glimpse of the implications that applying the BIOTREAT technology would have on a typi-cal Danish waterworks, in terms of GHG emissions is provided. The question we try to answer was how much would the carbon footprint of drinking water production increase by applying the con-cepts developed in BIOTREAT. In order to do this we consider the life cycle impact of a waterworks (excluding distribution to the consumer) and add on top of it the impact of either BIOTREAT or BI-OTREAT-carrier (sand). The result can be seen in deliverable 8.16 figure 40.

This figure shows that the BIOTREAT concept would have a rather small impact as far as drinking water production is concerned. BIOTREAT would increase GHG emissions by less than 4% whereas the use of carriers would lead to an increase of above 5% over current emissions. Most of the im-pacts associated with drinking water production are related to the electricity used for operation, and the additional emissions associated with BIOTREAT are comparatively small.

The following overall conclusions of the LCIA can be drawn:

• The environmental impact of applying the BIOTREAT technology is related to the produc-tion of the bacterial inoculum and to the energy used by the additional UV system that needs to be installed in the waterworks.
• When compared to GAC adsorption, BIOTREAT has an impact of a similar magnitude, with the exception of land use and aquatic eutrophication, where its performance is poorer. In the other impact indicators the actual scores are very often lower than GAC, implying a lower impact, but this cannot be taken at face value, given the uncertainty involved in the study.
• The performance of BIOTREAT is poorer in all impact indicators when compared to the al-ternative of well re-location. However this is only the case if we assume that the re-located wells will have the same pumping energy needs than current ones.
• Inoculation frequency can be increased without jeopardizing the environmental perfor-mance. This is true as long as bacteria are produced at large-scale, with efficiency similar to that of industrial yeast production.
• If the inoculum were produced at a small scale and inefficiently, as done in the pilot plant tested near Copenhagen, the environmental impact of BIOTREAT would be much higher than that of GAC. It is therefore of the utmost importance to make sure bacteria are pro-duced efficiently.
• Applying the BIOTREAT concept in a Danish waterworks would increase the carbon foot-print of drinking water production by less than 5%. If water distribution to consumers were accounted for, this percentage would be even lower.

• The use of carriers increases the impact of the BIOTREAT technology, due to the additional materials needed: mineral carrier, sodium alginate and calcium chloride. The impact of producing these materials and replacing them on an annual basis does not compensate for the reduced inoculation frequency. As we have seen, producing the inoculum involves a relatively low impact if done at industrial scale.
• The impact of this option is very sensitive to the inoculation frequency, with the impact quickly increasing if augmentation is required more than once per year.
• But it must be mentioned that if augmentation is kept to once per year, and instead of a high-value material it is decided to use sand mined in the waterworks area, the impact of using carriers is substantially reduced, making it competitive – in environmental terms –with GAC adsorption.
• The point above, added to the fact that the data used to assess carriers was insufficient (from small lab-scale experiments), prevents us from advocating against carriers use. Fur-ther research is encouraged in order to get a better picture of its feasibility and benefits, especially if it turns out to be the only way of stabilizing the bacteria in the filter.
• Applying BIOTREAT with carriers in a Danish waterworks would increase the carbon foot-print of drinking water production by around 5-6%, if an optimistic scenario is considered. If water distribution to consumers were accounted for, this percentage would be even lower.

Potential Impact:
Millions of tonnes of organic xenobiotics are used each year worldwide, for example the pesticides used in agricultural production and applied to paved urban areas, along railways and roads, and to farmyards. Pharmaceuticals are considered to be emerging contaminants that enter the environ-ment via the application of manure and sewage sludge to farmland, via effluent from wastewater treatment plants and via accidental spillage during various industrial applications. As a result of these extensive environmental inputs, European water bodies such as rivers, lakes and groundwa-ter that are used as drinking water resources have become contaminated with a wide range of organic micropollutants. Clean drinking water is a limited resource, not only in Third World countries but also in many European regions. It is becoming increasingly difficult to meet the quality standards of the European Drinking Water Directive regarding chemical residues of pesticides and other micropollutants and many potential drinking water resources have been abandoned due to exceedance of the EU limit values for these micropollutants. The EU limit value for individual pesticides in drinking water is 0.1 µg/l, while that for multiple pesticides is 0.5 µg/l (Directive 98/83/EC). These concentrations are very low, and many water bodies have therefore had to be abandoned as drinking water resources (Stockmarr, 2005). There is therefore an urgent need to develop new sustainable water treatment technologies that satisfy the EU quality standards for drinking water. The aim of BIOTREAT was to develop new sustainable treatment technologies that exploit the potential of microorganisms to mineralise a range of pollutants without the accumulation of unwanted degradation products. The impacts of the project on water supply and science are discussed below.
4.1 SCIENTIFIC IMPACTS
The incredible capacity of soil and subsurface microbial communities to degrade a wide range of xenobiotic compounds is well known. In most cases where microorganisms are used for bioreme-diation, however, the biodegradation processes operate as “black box” systems, and detailed knowledge of the microorganisms and the metabolic processes and pathways involved is lacking. As complete mineralisation of organic micropollutants may require specialised strains that are rarely present in the indigenous microbial community (e.g. Topp et al., 2000; Sørensen et al., 2007), bioaugmentation with degradative bacteria has been proposed as a strategy for remedia-tion of contaminated drinking water resources.

BIOTREAT has provided essential knowledge opening the “black box” of bioremediation of drinking water polluted by micropollutants. Furthermore, the project has filled the gap between laboratory studies and full-scale application. Selected main findings are listed below:
• The finding of a hitherto unknown potential for degradation of micropollutants, including bentazone, MCPA and BAM, but only in in certain waterworks sand filters.
• Development of Low Flux Filters to be used for isolation of bacteria degrading contami-nants at trace concentrations.
• Complete genetic characterization of the BAM degrading bacterium Aminobacter MSH1, including biochemical characterization of the BAM-degrading amidase bbdA gene.
• Development of techniques for immobilization of degrader bacteria on specific carriers to be used for bioaugmentation of waterworks sand filters to improve degrading efficiency.
• Determining the effect of assimilabe organic carbon on degradation of target micropollu-tants.
• Elucidation of the carbamazepine degradation pathway in sewage sludge and waterworks sand filters, including accumulation of recalcitrant transformation products.
• Identification of protozoa in waterworks sand filters and clarification of their role as preda-tors feeding on introduced bacteria.
• Development of tools for interpretation of molecular fingerprinting pattern (e.g. PCR-DGGE), including the range-weighted richness (Rr), dynamics (Dy) and functional organiza-tion (Fo) parameters for the different experimental setups.
• The development of a model framework to describe growth-linked biodegradation of trace-level pollutants in the presence of coincidental carbon substrates and microbes.
• The development of a continuum biofilm model to describe metabolic and cometabolic trace pollutant removal scenarios in waterworks sand filter systems.
• Demonstration of the BIOTREAT metabolic strategy in pilot-scale sand filters inoculated with the BAM degrading Aminobacter bacterium.
• Defining degrading kinetics constrains limiting the cometabolic degradation strategy at pi-lot-scale systems simulating waterworks sand filters.
• The development of a prospective environmental and economic assessment for biotreat-ment of micropollutants in drinking water resources.
These main scientific findings have all been published in high-ranking international journals with peer review or are submitted or in preparation for publication. All together 39 scientific articles have been published, 2 have been submitted, 10 are in preparation for publication and several more are foreseen. In addition the findings have been disseminated at international conferences, symposia and workshops.
4.2 OUTPUT TO STAKEHOLDERS, MANAGERS AND EXPERTS WITHIN THE WATER SUPPLY
Within BIOTREAT, new and much needed sustainable biotechnologies have been developed for remediating contaminated water from subterranean and surface drinking water resources. The technologies focus on remediation of waters polluted by organic chemicals at trace pollutant con-centrations typically at the µg to ng range. The basis of the proposed technologies is bioaugmenta-tion, which in the present context is the introduction of specific degrading microorganisms or mi-crobial consortia into existing sand filters at waterworks. The basic idea is that once the technolo-gy has been developed for waterworks sand filters it can easily be transferred also to other types of submerged biofilter systems such as mobile biofilters placed close to groundwater abstraction wells, sand barriers between surface waters and abstraction wells and subterranean protective barriers established to prevent micropollutants from entering into aquifers.

Little is known about degradation at trace pollutant concentrations and bioaugmentation of wa-terworks sand filters, as a technology to remediate polluted drinking water has not been attempt-ed previously. BIOTREAT has exploited the potential of both metabolic and cometabolic processes to remediate water resources abandoned or in danger of being abandoned due to the presence of low concentrations of micropollutants and thereby enable them to meet the EU quality criteria for drinking water. Addition of degrader bacteria to waterworks sand filters as a technology to reme-diate drinking water polluted by trace concentrations is not trivial, but it is of great interest. Sever-al laboratory studies have shown that such technology may be feasible, but until now no one has striven to investigate this in pilot-scale simulating real waterworks sand filters. The challenges go-ing from laboratory-scale to full-scale are many. Firstly, the short retention time of sand filters leaves very short time for degradation to occur, secondly pollutant concentrations may be in the ng to µg/l range and thus too low to support growth of degrader bacteria at the prevailing flow conditions, and thirdly at real filter conditions loss of bacteria may occur either because of de-tachment of degrader bacteria due to the rapid flow across the filters, because of predation by native protozoa in the filters, or due to loss during waterworks operation procedures. These chal-lenges have all been experimentally addressed in BIOTREAT and suggestions for further improve-ment of the technology have been put forward.

In general the metabolic strategy showed to be the most effective. Pilot scale field studies examin-ing the potential for bioaugmentation with the BAM-degrading Aminobacter sp. MSH1 in sand fil-ters operated at the rapid flow conditions prevailing in waterworks filters resulted in up to 75% removal of 0.2 µg/l BAM, giving concentrations in the purified water below the 0.1 µg/l legal threshold limit. In addition, no BAM degradation products were observed and no adverse effects of inoculation on important filter processes such as oxidation of iron and ammonium were observed. Though, it was difficult to maintain efficient degradation for longer time periods due to loss of MSH1-bacteria which occurred mainly during backwashing operation. By avoiding backwash procedures, the degradation was prolonged, but bacteria, and hence degradation activity, were still lost with time.

The exploitation of the cometabolic strategy was less convincing. Although laboratory experiments showed cometabolic degradation of several trace pollutants this could not be demonstrated in a test pilot sand filter reactor. Based on this it was concluded that the specific process conditions in waterworks sand filters do not allow for co-metabolic conversion, at least not for the micropollu-tants investigated in the project.

The above research has been disseminated to a broader audience of stakeholders, managers and expert within the water supply at the BIOTREAT open-end users meetings, with partition of the BIOTREAT end-users board. At the meetings BIOTREAT partners gave presentations of the main findings followed by intensive discussions. At the first meeting which was held 4th December 2012 in Leuven, Belgium, the end-users still had the possibility to influence the project in relation to immediate needs in water supply. To example it was emphasized by the end-users to carry out ex-periments at real trace pollutant concentrations and at conditions simulating real waterworks op-erational procedures, including realistic hydraulic retention times and backwashing. The end-users were also asked to contribute to the selection of model pollutants of practical relevance in Euro-pean water supply. The second open-end user meeting were the main outcome of the project were given was held 4th November 2014 in Copenhagen, Denmark. At this meeting there was a thorough discussion with the end-users about next steps to be taken for further exploitation of the developed technologies in practice.

The BIOTREAT technologies has also been disseminated to a broader audience of stakeholders, managers, experts and scientists involved in bioremediation and biotechnology in general at the following meetings:

• 5th European Bioremediation Conference, 4-7th July Chania, Crete, Greece. (Special session about KBBE)
• Environmental Microbiology and Biotechnology – In the frame of the knowledge-based bio and green economy (EMB2012) 10-12th April 2012, Bologna, Italy
• “Bioremediatie van pesticiden” Leuven, Belgium 27/3/2014. A mini-symposium on pesti-cide biodegradation for Flemish stakeholders including talks on general biodegradation of pesticides and results from the BIOTREAT project.
A technology action plan including a business plans has been developed for 1) The BIOTREAT bio-augmentation technology focusing on remediation of BAM in waterworks sand filters and 2) the BIOTREAT carrier technology for immobilization of degrader bacteria. The technology action plan identifies the main competing technologies being granular activated carbon (GAC) treatment or more simple well relocation. The latter is in many cases a more obvious solution, but is often not an option because clean drinking water resources may not be available in the vicinity. A compari-son of the operating costs of the BIOTREAT techniques with the operating costs of the traditional GAC technology shows that the BIOTREAT techniques are mainly financially interesting for small water suppliers.

Three steps have been envisaged in the process towards the development of a commercial proto-type where the funding of BIOTREAT has covered the pre-competitive phase. The next step will be the innovative phase involving further development of the technology focusing on more competi-tive issues. The innovative phase will include large-scale demonstration of the technology empha-sizing specific applications for the treatment of contaminated water using e.g. 1) existing sand fil-ters at waterworks 2) mobile biofilters placed close to groundwater abstraction wells, 3) sand bar-riers between surface waters and abstraction wells and 4) subterranean protective barriers estab-lished to prevent micropollutants from entering into aquifers. The innovative phase also includes a survey of regulatory and legislative issues related to the use of microorganisms in water treatment processes. Funding for this phase may be sought through the EUREKA network for market-oriented R&D, and/or private sources. The final step will be Commercialisation where specific bioaugmentation applications and encapsulation technologies are launched on the market. It is currently discussed by the BIOTREAT partners how to go from the pre-competitive phase to the innovative phase.
4.3 OUTPUT TO THE GENERAL AUDIENCE
The general public has been informed about the project through the BIOTREAT homepage (www.biotreat-eu.org) and through a project leaflet which was distributed broadly at different meetings and events. Both the homepage and the leaflet give broad introduction to the project.
4.4 EDUCATION OF EARLY STAGE RESEARCHERS
The education of a new generation of environmental scientists able to meet problems of tomor-row concerning the increased pollution of water resources by organic chemicals is urgent. Several early stage researchers have been educated with funding from BIOTREAT including both PhDs and postdoc fellows. BIOTREAT has therefore contributed to the education of the next generation en-vironmentalists having specific expertise in bioremediation technologies. Three young fellow workshops have been held within BIOTREAT to improve teamwork and to provide a long lasting network for the fellows also beyond BIOTREAT. The workshops were arranged by the fellows themselves and included sessions about knowledge sharing, dissemination and improvement of presentation skills.


CONCLUSIONS

The multidisciplinary BIOTREAT consortium has been working for more than 48 month on the de-velopment of new water treatment technologies to be used for remediation of drinking water pol-luted by pesticides, pharmaceuticals and other micropollutants. Two remediation strategies have been examined; a strategy based on metabolic processes for micropollutants where metabolically degrading bacteria were available and a cometabolic strategy, relying on unspecific degradation by methane or ammonium oxidizing bacteria, for target pollutants where specific degrading bacteria have not been described. The results of the metabolic strategy was most promising as up to 75% of the investigated pollutant (BAM) was degraded at realistic waterworks flow conditions with a hydraulic retention time of about 20 minutes. The technology may be improved by measures preventing loss of degrader bacteria from the filter as it was difficult to maintain effi-cient degradation for longer time periods. Immobilisation of degrader bacteria on carriers was a promising technology as it diminished loss of degrader bacteria from the filter and thereby pro-longed the period of effective degradation. The cometabolic strategy was less promising as no mi-cropollutants removal was observed going from laboratory scale experiments to large-scale filters operated at waterworks sand filter conditions. Cost-benefit analysis and life cycle impact assess-ments showed that the BIOTREAT metabolic strategy was competitive to granular activated car-bon treatment being the most likely competitor technology to the BIOTREAT technology. Both cost and environmental impact increased using the BIOTREAT carriers in combination with the BIO-TREAT metabolic remediation strategies. However, cost and environmental impact could possibly be lowered selecting other carrier materials as to example ordinary quartz sand.


REFERENCES

Albers, C. N., L. Feld, L. Ellegaard-Jensen, J.Aamand (2015) Degradation of trace concentrations of the per-sistent groundwater pollutant 2,6-dichlorobenzamide (BAM) in bioaugmented rapid sand filters (Submitted to Water Research)
Begonja, A and D. Hrsak (2001) Effect of growth conditions on the expression of soluble methane monooxygenase. Food Tech. Biotechnol.39:29-35
Chavez de Paz, L. E., and A. Resin (2011). Antimicrobial effect of chitosan nanoparticles on streptococcus mutans biofilms. Appl Environ Microbiol 77: 3892-3895.
Choi, D.W. R.C Kunz, E.S. Boyd, J. D. Semrau, W.E Antholine, J.I. Han, J.A. Zahn, J. M. Boyd, A.M. de la Mora, A.A DiSpirito (2003) The membrane-associated methane monooxygenase (pMMO) and pMMO-NADH : quinone oxidoreductase complex from Methylococcus capsulatus bath. Journal of bacteriology 19:5755-5764
Daims, H., S. Lücker, and M. Wagner (2006) Daime, a novel image analysis program for microbial ecology and biofilm research. Environ. Microbiol. 8: 200-213
Egli, T. (2010). "How to live at very low substrate concentration." Water Research 44(17): 4826-4837
Han, J. I., S. Lontoh, and J. D. Semrau (1999) Degradation of chlorinated and brominated hydrocarbons by Methylomicrobium album BG8. Arch. Microbiol. 172:393-400
Hanson, R. S. and T. E. Hanson (1996) Methanotrophic bacteria. Microbiol Rev. 60: 439–471
Heydorn, A., A.T. Nielsen, M. Hentzer, C. Sternberg, M. Givskov, B. K.Ersboll,and S. Molin, S (2000) Quantification of biofilm structures by the novel computer program COMSTAT. Microbiology-UK 146: 2395-2407
Ho, A., F-M. Kerckhof, C. Luke, A. Reim, S. Krause, N. Boon and P. L. E. Bodelier (2013) Conceptualizing functional traits and ecological characteristics of methane-oxidizing bacteria as life strategies. Environ. Microbiol. Report 5:335–345
Lee, S.W. D R. Keeney, D-H Lim, A. A. Dispirito, and J. D. Semrau (2006) Mixed pollutant degradation by Methylosinus trichosporium OB3b expressing either soluble or particulate methane monooxygenase: Can the tortoise beat the hare?
Lontoh, S., and J. D. Semrau (1998) Methane and trichloroethylene degradation by Methylosinus tricho-sporium OB3b expressing particulate methane monooxygenase. Appl. Environ. Microbiol. 64:1106-1114
Lontoh, S., A. A. DiSpirito, and J. D. Semrau (1999) Dichloromethane and trichloroethylene inhibition of methane oxidation by the membrane-associated methane monooxygenase of Methylosinus trichosporium OB3b. Arch. Microbiol. 171:301-308
Mueller, L. N., and J. F. de Brouwer (2006). Analysis of a marine phototrophic biofilm by confocal laser scanning microscopy using the new image quantification software PHLIP. BMC Ecol 6: 1
Oldenhuis, R., J. Y. Oedzes, J. J. Van der Waarde, and D. B. Janssen (1991) Kinetics of chlorinated hydrocar-bon degradation by Methylosinus trichosporium OB3b and toxicity of trichloroethylene. Appl. Environ. Microbiol. 57:7-14
Oldenhuis, R., R. L. Vink, D. B. Janssen, and B. Witholt (1989) Degradation of chlorinated aliphatic hydrocarbons by Methylosinus trichosporium OB3b expressing soluble methane monooxygenase. Appl. Environ. Microbiol. 55:2819-2826
Schwarzenbach, R.P. B. I. Escher, K. Fenner, T. B. Hofstetter, C. A. Johnsen, U. von Gunten, and B. Wehrli (2006) The challenge of micropollutants in aquatic systems. Science 313:1072-1077
Stockmarr, J. (2005) Groundwater quality monitoring in Denmark. Geological Survey of Denmark and Greenland Bulletin. 7: 33–36
Sørensen, S. R., M. S. Holtze, A. Simonsen, and J. Aamand (2007) Degradation and mineralization of nano-molar concentrations of the herbicide dichlobenil and its persistent metabolite 2,6-dichlorobenzamide by Aminobacter spp. isolated from dichlobenil-treated soils. App.Environ. Microbiol. 73: 399-406
Topp, E., H. Zhu, S. M. Nour, S. Hout, M. Lewis, and D. Cuppels (2000) Characterization of an atrazine-degrading Pseudaminobacter sp. Isolated from Canadian and French agricultural soils. Appl. Environ. Mi-crobiol. 66: 2773-2783.


List of Websites:
Project website: www.biotreat-eu.org

List of participants: List of participants:

1.Geological Survey of Denmark and Greenland (GEUS), Denmark. Prof Jens Aamand (jeaa@geus.dk)
2.Danmarks Tekniske Universitet (DTU), Denmark. Prof Barth Smets (bfsm@env.dtu.dk)
3.Katholieke Universiteit Leuven (KU LEUVEN), Belgium. Prof Dirk Springael (Dirk.springael@ees.kuleuven.be)
4.Eidgenoessishe Anstalt fur Wasserversorgung Abwasserreinigung und Gewassershutz (EAWAG), Switzerland. Dr Hans-Peter E. Kohler (hkohler@eawag.ch)
5.Universiteit Gent (UGENT), Belgium Prof Nico Boon (Nico.boon@ugent.be)
6.Bundesanstalt für Gewässerkunde (BfG) PD Dr Thomas Ternes, Germany (ternes@bafg.de)
7.Institut za microbioloske znanostiin Tehnologije DOO (IMST) Robert Ravnihar, Slovenia (robert.ravnihar@gmail.com)
8.BIOCLEAR B.V.(BIOCLEAR) Netherland. Marlea Wagelmans (wagelmans@bioclear.nl).
9.AVECOM N.V. (AVECOM), Belgium. Mariane Van Wambeke (mariane.vanwambeke@avecom.be)
10.2.-0 LCA consultants ApS (2.-0 LCA), Denmark. Randi Dalgaard (rad@lca-net.com)
11.Vlaamse maatschappij voorwatervoorziening cvba (VMW) Belgium, Julie Degryse (julie.degryse@dewatergroep.be)