Chemosphere 363 (2024) 142956 Available online 27 July 2024 0045-6535/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). How bioaugmentation for pesticide removal influences the microbial community in biologically active sand filters Laura Pickering a, Victor Castro-Gutierrez b, Barrie Holden c, John Haley c, Peter Jarvis a, Pablo Campo a, Francis Hassard a,* a Cranfield University, College Road, Cranfield, Bedfordshire, MK43 0AL, UK b Environmental Pollution Research Center (CICA), University of Costa Rica, Montes de Oca, 11501, Costa Rica c UK Water Industry Research Limited, London, UK H I G H L I G H T S G R A P H I C A L A B S T R A C T • Augmented Sphingobium ensured 15-day metaldehyde water compliance in SSF. • Sphingobium abundance declined quickly after dosing. • Sphingobium metaldehyde removal suc- cess tied to its distribution, persistence, and activity. • Bioaugmented agents’ impact on microbiome varied by species in filter. • For optimal pesticide removal, enduring and active bioaugmented bacteria are key. A R T I C L E I N F O Handling editor: A ADALBERTO NOYOLA Keywords: Metaldehyde Micropollutant Drinking water Slow sand filter Water treatment A B S T R A C T Removing pesticides from biological drinking water filters is challenging due to the difficulty in activating pesticide-degrading bacteria within the filters. Bioaugmented bacteria can alter the filter’s microbiome, affecting its performance either positively or negatively, depending on the bacteria used and their interaction with native microbes. We demonstrate that adding specific bacteria strains can effectively remove recalcitrant pesticides, like metaldehyde, yielding compliance to regulatory standards for an extended period. Our experiments revealed that the Sphingobium CMET-H strain was particularly effective, consistently reducing metaldehyde concentrations to levels within regulatory compliance, significantly outperforming Acinetobacter calcoaceticus E1. This success is attributed to the superior acclimation and distribution of the Sphingobium strain within the filter bed, facilitating more efficient interactions with and degradation of the pesticide, even when present at lower population den- sities compared to Acinetobacter calcoaceticus E1. Furthermore, our study demonstrates that the addition of pesticide-degrading strains significantly impacts the filter’s microbiome at various depths, despite these strains * Corresponding author. E-mail address: francis.hassard@cranfield.ac.uk (F. Hassard). Contents lists available at ScienceDirect Chemosphere journal homepage: www.elsevier.com/locate/chemosphere https://doi.org/10.1016/j.chemosphere.2024.142956 Received 7 March 2024; Received in revised form 10 July 2024; Accepted 25 July 2024 mailto:francis.hassard@cranfield.ac.uk www.sciencedirect.com/science/journal/00456535 https://www.elsevier.com/locate/chemosphere https://doi.org/10.1016/j.chemosphere.2024.142956 https://doi.org/10.1016/j.chemosphere.2024.142956 https://doi.org/10.1016/j.chemosphere.2024.142956 http://crossmark.crossref.org/dialog/?doi=10.1016/j.chemosphere.2024.142956&domain=pdf http://creativecommons.org/licenses/by/4.0/ Chemosphere 363 (2024) 142956 2 making up less than 1% of the total microbial community. The sequence in which these bacteria are introduced influences the system’s ability to degrade pesticides effectively. This research shows the potential of carefully selected and dosed bioaugmented bacteria to improve the pesticide removal capabilities of water filtration systems, while also highlighting the dynamics between bioaugmented and native microbial communities. Further investigation into optimizing bioaugmentation strategies is suggested to enhance the resilience and efficiency of drinking water treatment systems against pesticide contamination. 1. Introduction The Food and Agriculture Organization reports 2.5–4.1 million metric tons of pesticides applied globally, yet less than 25% actually reach their target organisms, with the majority ending up in the envi- ronment through leaching and run-off (Pretty and Bharucha, 2015; Pretty, 2018). This widespread use of pesticides poses significant risks to ecosystems and human health, through ingestion of contaminated food or water, inhalation of vapours or particles, and dermal exposure to treated or contaminated areas. Pesticides undergo transformation via biogenic and abiotic pathways, creating various residues, some of which are harmful or recalcitrant and can persist for decades, causing long-term environmental impacts (Fenner et al., 2013). Their high mobility in soil often leads to drinking water contamination, and con- ventional water treatment technologies struggle to remove certain polar and/or low molecular weight compounds (Rolph et al., 2018). Climate change, with its irregular and intense rainfall patterns, further amplifies pesticide mobilization into surface and groundwater, increasing the need for sustained and higher application (Schönenberger et al., 2022; Toccalino et al., 2014). The ongoing detection of stable transformation products from both current and phased-out pesticides in groundwater highlights the enduring nature of these pollutants (Fenner et al., 2013; Gilliom et al., 1992). To safeguard public health, national regulatory bodies have estab- lished conservative, health-based standards for acceptable contaminant concentrations in drinking water. For instance, in Europe (including UK), the maximum allowable concentration for total pesticides and their transformation products is set at <0.5 μg L− 1, while individual com- pounds are limited to <0.1 μg L− 1 (Council Directive 98/83/EC, 2010). In the UK, mecoprop, propyzamide, and metaldehyde accounted for approximately 30% of drinking water chemical compliance failures in 2020 (Drinking Water Inspectorate, 2021), underscoring their signifi- cance as a chemical exposure pathways through drinking water. Pesti- cide removal methods include adsorption, membrane filtration (reverse osmosis, RO, nanofiltration), chemical oxidation (ozonation, chlorina- tion), and UV photodegradation, but these can be costly, especially as retrofit or ‘polishing’ treatments (Rolph et al., 2018, 2020). Cost-effective alternatives like catchment interventions and safer prod- uct subsidies show promise (UKWIR, 2015; Mohamad Ibrahim et al., 2019; Balashova et al., 2021). Biological drinking water filters are treatment systems that use natural microbial communities to remove contaminants and impurities from water through biological processes. However, the most applied treatment for micropollutants is Granular activated carbon (GAC) which can remove around 70% of pesticides, yet some adsorption resistant compounds evade filtration, compromising water safety (Taylor et al., 2022). Biodegradation in catchments is the primary route to break down many pesticides, highlighting the potential for research into biological treatments that leverage microbial commu- nities for contaminant removal in water systems (Alexander et al., 2013; Hassard et al., 2016; Li et al., 2022). Pesticides can alter microbial community structures and functions, including the evolution of new degradation enzymes and pathways (Haig et al., 2015; Castro-Gutiérrez et al., 2020). These capabilities, often shared among microbes via horizontal gene transfer (Zhang et al., 2006), enable enhanced biodegradation with repeated exposure (Simms et al., 2006). Yet, pinpointing these adaptive mechanisms in microbes is difficult due to pesticides’ transient presence, complex ecological influences, and the varied evolution of degrading functions across mi- crobial species (Castro-Gutiérrez et al., 2020). Given that pesticides appear in ng-μg levels in waters, this often results in the slow growth of degraders or the selection of inefficient pathways in treatment settings – possibly resulting in intermediates or byproducts. This underscores the importance of innovative strategies to utilize these microbial processes for improved degradation rates in environmental and water treatment systems (Castro-Gutierrez et al. 2022; Cosgrove et al., 2022; Zhang et al., 2022). To enhance biodegradation in biological systems, biostimulation and bioenrichment are applied. Biostimulation involves adding nutrients or electron acceptors to enhance biomass growth for better biodegradation or enzyme production (Aldas-Vargas et al., 2021), while bioenrichment introduces specific pollutants to promote the growth of degrading mi- crobes (Wang et al., 2020). In drinking water treatment, these ap- proaches can be used in side-stream setups, but have rarely been implemented at full-scale (Rolph et al., 2019). In water treatment, where microbial activity often decreases due to predation, washout, or nutrient scarcity (Pérez et al., 2016; Horemans et al., 2017), bioaugmentation offers a solution by adding specific degraders to increase degradation efficiency (Albers et al., 2015; Ma et al., 2022). This method involves introducing external microorganisms, grown ex situ, including bacteria with degradation pathways or plasmids, and/or mobile genetic elements that spread degrading genes (Dutta et al., 2022; Rios Miguel et al., 2020). The challenge is to ensure these introduced organisms or genetic elements are effectively established and functional within the target environment, remain contained without compromising drinking water safety (Vandermaesen et al., 2022; Pinilla-Redondo et al., 2021). Previous research has explored bioaugmentation in water filters to remove specific pesticides, observing variable treatment effectiveness due to factors like bacterial adhesion or interactions with resident mi- crobes (Albers et al., 2015; Bouchez et al., 2000; McDowall et al., 2009; Samuelsen et al., 2017). Enhancing the stability of Aminobacter sp. MSH1 in filters, through immobilization in sand or biocarriers, has been shown to sustain removal rates of specific micropollutants like 2, 6-dichlorobenzamide (BAM) (Albers et al., 2014; Horemans et al., 2017). Schostag et al. (2022) combined RO with microbial degradation for BAM-contaminated water, finding RO effectively pre-concentrated pesticides for more efficient microbial degradation. However, chal- lenges like low contact time between degraders and pollutants can limit removal efficiency (McDowall et al., 2009) and performance can depend on the adequate presence and distribution of degraders in the system (Castro-Gutiérrez et al., 2022b). While bioaugmentation’s impact on native microbial communities is generally minimal (Castro-Gutiérrez et al., 2022b; Schostag et al., 2022), interactions with indigenous mi- crobes in filters can vary, affecting the survival of introduced degraders (Vandermaesen et al., 2022). Artificially elevating pollutant loads can aid in establishing specific degraders populations but pose risks of supply contamination and secondary pollution, thus making it a less favoured option for water utilities. One pesticide receiving research attention is metaldehyde is widely used in agriculture as a molluscicide to control slugs and snails, especially in cereal, vegetable, and orna- mental crops. It works by disrupting mucous production, leading to dehydration and death in these pests. Metaldehyde is toxic if ingested, causing symptoms like nausea, vomiting, diarrhea, and in severe cases, seizures or death. It poses risks to non-target organisms and can contaminate water sources, impacting aquatic life and wildlife. L. Pickering et al. Chemosphere 363 (2024) 142956 3 This study suggests that adding bioaugmentation agents for metal- dehyde removal in drinking water systems impacts microbiome di- versity, potentially influencing micropollutant elimination (Haig et al., 2015). The most pronounced effects are expected close to the dosing area, as the competition for resources within this oligotrophic environ- ment could affect the persistence of bioaugmented strains, but ecolog- ical niches could establish in other locations of the filter. Such competitive interactions might modify the metabolic activities, degra- dation potentials, and symbiotic relationships of both native and augmented microbes, thereby affecting their ability to effectively remove the target pesticide and alter the augmentation’s impact on water treatment reactor functionality (Sun and Jing, 2023). This research aim to investigate the impact of bioaugmentation on pesticide removal efficacy andmicrobial community dynamics in slow sand filters (SSF). The scientific novelty of this research lies in the demonstration of how bioaugmentation with specific bacterial strains influences the mi- crobial community structure and function in biologically active sand filters for pesticide removal. This study highlights the impact of bio- augmentation sequencing and the interaction between augmented and native microbial communities, providing insights into the mechanisms driving enhanced pesticide degradation and the importance of main- taining microbiome integrity for sustained water treatment. Under- standing degrader distribution and competition in plug flow reactors is essential for developing more efficient biological filtration systems for micropollutant removal, enhancing reactor design, bioaugmentation strategies, ensuring optimal performance and ultimately regulatory compliance. 2. Materials and methods 2.1. Chemicals and water for pilot sand filter experiments Metaldehyde is a cyclic tetramer of acetaldehyde with a molecular mass of about 176.21 g/mol. It exhibits low water solubility, low vapor pressure, and a moderate lipophilicity, with a log Kow of 1.97, indicating its affinity for organic material. Its adsorption coefficient (Koc) ranges from 210 to 1800, suggesting varying potential for soil binding depending on environmental conditions. In terms of chemical acquisi- tion, methanol (≥99.6 %) was purchased from Fisher Scientific (UK), and metaldehyde (>99%) from Acros Organics (NJ, USA). To minimize the risk of cross-contamination of the target pesticide, all equipment was thoroughly cleaned using 99% acetone and stored in a fume hood overnight. For dosing the pilot-scale columns, partially treated water was sampled from aWater Treatment Works (WTW) facility operated by a UK water utility, situated in the southern region of England, UK. This water sample was representative of the type typically subjected to SSF, as it was collected directly before entering a full-scale SSF. The water from the River Thames had undergone multiple treatment stages, including reservoir storage, treatment, coagulation-flocculation, and direct depth filtration via a Rapid Gravity Filter (RGF). 20 m3 of water was transported in several batches to the UK National Research Facility in Water and Wastewater Treatment at Cranfield University. The ex- periments were conducted within 3 months of the water’s delivery. To prepare the water samples for testing in continuous flow pilot- scale SSF experiments, metaldehyde was added to 1000 L batches. A metaldehyde solution with a concentration of 10 mg L− 1 was created by dissolving analytical grade metaldehyde (>99%, Thermofisher, UK) in ultrapure water (PureLab Option s7/15, 18.2 MΩ cm, and TOC <3 ng L− 1). A 0.2 L aliquot of this solution was then added to each batch to achieve a final concentration of 2 μg L− 1, which is representative of realistic environmental concentrations reported in source waters in the UK and other regions (Balashova et al., 2021). DOC and the following nutrients NH4–N, NO3–N were analysed using standard methods 5310B, 4500-NH3 G and 4500-NO3 F from 0.45 μm filtered water, which was stored at 4 ◦C before analysis. Turbidity measurements were performed on unfiltered water according to standard method 2130B (APHA, 2023). The inlet dissolved organic carbon was 4.21–5.85 mg L− 1, the UV254 was 0.026–0.063 cm− 1 and the turbidity was 0.305–1.01 nephelometric turbidity units (NTU). In terms of water chemistry, the pH was 8.09–8.64, and NO3–N was 0.7–1.0 mg L− 1. There was no detectable NH4–N in the inlet water. 2.2. Analytical methods for metaldehyde Metaldehyde was determined by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Analyte separation was conducted using an ExionLC Series UHPLC (Sciex, USA). The analytical column was an Acquity UPLC BEH C18 column (1.7 μm, 2.1 × 50 mm) from Waters (USA), which was operated at 60 ◦C. The mobile phase consisted of water and methanol; both eluents were buffered with 2 mM ammonium formate. The initial composition of the elution gradient was set at 95% water. The methanol concentration was linearly increased to 95% within 2.4 min. Subsequently, the gradient composition was held for 0.1 min; then the column was re-equilibrated to the initial gradient composition for 2.5 min. Aqueous samples (10 μl) were directly injected into the chromatograph. Metaldehyde was ionized in positive mode in a SCIEX QTRAP 6500+ (Sciex, USA); the electrospray parameters included: capillary voltage 5.5 kV; curtain gas, 30 psi; nebulizer gas 70 psi, drying gas, 40 psi; gas temperature, 200 ◦C. The ammonium adduct of metaldehyde (m/z 194) was monitored, and its product ions atm/z 62 and 106 were used for quantitation and verification purposes, respec- tively; the corresponding declustering potential and collision energies were 11 and 9 V. The external calibration curve, comprising five con- centration levels, showed excellent linearity (r2 > 0.999) and accuracy (±5%) between 0.1 and 20 μg L− 1; the detection limit was set at 0.01 μg L− 1. 2.3. Bacterial strains used for bioaugmentation Acinetobacter calcoaceticus E1 and Sphingobium CMET-H was isolated according to Castro-Gutiérrez et al. (2020) and Castro-Gutiérrez et al. (2022b) respectively. Acinetobacter calcoaceticus E1 and Sphingobium CMET-H were chosen due to their previously demonstrated capabilities in degrading metaldehyde. Both strains were isolated from contami- nated soil environments known for their metaldehyde exposure. Acine- tobacter calcoaceticus E1 is a Gram-negative bacterium often isolated from hydrocarbon-contaminated environments. It is known for its ver- satile metabolism, particularly its ability to degrade various hydrocar- bons and aromatic compounds, making it suitable for bioaugmentation in water treatment systems due to its capability to break down organic pollutants. Sphingobium CMET-H is a Gram-negative bacterium isolated from pesticide-contaminated environments. It specializes in degrading complex organic pollutants, including metaldehyde, through specific metabolic pathways encoded on plasmids. To precisely dose degraders into the reactor, calibration curves for OD600 nm growth in Luria Bertani (LB) broth during the exponential phase were established and validated using a plate count method and flow cytometry to quantify the cultur- able number and total cell count. Bioaugmentation with A. calcoaceticus E1 at 1 × concentration (2.40 × 107 CFU mL− 1) and 2 × concentration (4.84 × 107 CFU mL− 1) was carried out on days 16 and 42, respectively, for filters 2, 3, and 4. These dosing events corresponded to Phase 1 of the experimental programme (Table 1). Due to suboptimal metaldehyde removal at the beginning of Phase 1, additional bioaugmentation dosing was performed during Phase 2. A. calcoaceticus E1 at roughly 3× the original dosed concentration (8.11 × 107 CFU mL− 1) was introduced on day 56 in filter 3, while Sphin- gobium CMET-H at a concentration of 5.00 × 107 CFU mL− 1 was dosed on day 56 in filters 4 and 5. These dosing events represented Phase 2 of the trial (Table 1). To assess the abundance of the strains, serial dilution plate counts were conducted on the inoculum using Luria-Bertani (LB) agar, taking into account the wet volume and porosity of the bed without biomass. L. Pickering et al. Chemosphere 363 (2024) 142956 4 2.4. Trials of bioaugmentation strains in pilot-scale flow through SSF Six Perspex filter columns with covers were constructed, each having a height of 1.29 m, an internal diameter of 0.15 m, and a volume of 22.8 L. The pilot SSF systems were filled with a sand bed atop a gravel layer with the following particle size distribution (PSD) and uniformity co- efficient (Cu) properties: 0.8 m sand layer (PSD: 0.1–0.3 mm, Cu: 1.35) on a 0.2 m gravel support media (PSD: 1–5 mm, Cu: 1.35). Sand was sourced from Specialist Aggregates Ltd. (UK) and was comprised of silica sand 1–2 mm (BS 8:16) with the following characteristics: The compo- sition includes 92.6% SiO2, 0.8% Al2O3, less than 6.46% Fe2O3, and a loss on ignition of 0.64%. Sand was washed thrice prior to use in col- umns with potable water. For testing, the sub-potable water was pum- ped to each filter column at a flow rate of 1.8 L h− 1 (0.102 m/h) using a peristaltic pump (530 U, Watson Marlow, UK). Two SSF columns functioned without metaldehyde or bio- augmentation agent dosing, serving as experimental controls (Table 1). The pilot SSFs had an empty bed contact time (EBCT) of 9.5 h and a hydraulic retention time (HRT) of 3.5 h. The experimental setup is illustrated in Fig. 1, and the operational design is shown in Table 2. Water samples were collected three times per week to monitor param- eters and water quality throughout the 72-day experiment. Skimming, a mechanical intervention in SSF was not undertaken in this trial due to the low particulate load in inlet water and the relatively short duration of the trial – which was within the range of filter run times for full scale SSF of 30–90 days. To examine the impact of bioaugmentation agents and pesticide doses on the microbial community, surface sand/biofilm samples were taken intermittently on days 1, 16, 18, 23, 30, 37, 42, 44, 51, 56, 58, 65, and 72. The selection of 13 intervals was to ensure comprehensive temporal coverage for all six columns. At the conclusion of the experiment (day 72), the biofilters were disassembled to allow sand samples to be extracted at depths of 0.1 m and 0.2 m from the top of the bed using dedicated sampling ports. Horizontal subsections were obtained with a sterile sand corer to quantify the number of pesticide degraders and their distribution throughout the cross-section (10 cm depth) of each SSF bed. 2.5. DNA extractions from SSF media DNA was extracted from the biofilm associated with 0.4 g of sand according to manufacturer’s instructions using the NucleoSpin Soil DNA extraction kit (Macherey-Nagel, Germany). The purity and concentra- tion of DNA extract was quantified using Jenway Genova Nano spec- trophotometer (Cole Parmer, UK) and Invitrogen Qubit 3 (Thermo Fisher Scientific, UK). DNA amplicons from the V4–V5 region of the 16s rRNA gene were sequenced using an MiSeq instrument (Illumina, USA). 2.6. 16S rRNA gene amplicon sequencing for microbial community analysis The V4–V5 region of the 16S rRNA gene was amplified from DNA using polymerase chain reaction, and DNA fragments were amplified with 515F/806R primers (Walters et al., 2016). The 16S rRNA ampli- cons were sequenced on a MiSeq (Illumina, USA) using 300 × 300 bp paired-end sequencing. Fastqc software (v. 0.11.9) was employed for quality checking of the amplicon sequence information, and QIIME2 (v. 2020.6.0) (Caporaso et al., 2010) was used for read sequence analysis. Demultiplexed paired-end sequences were imported, and the Divisive Amplicon Denoising Algorithm 2 (DADA2) was implemented for quality filtering (Q score ≥30) and chimera removal. Gene sequences with an abundance of <5 were excluded from the analysis after constructing amplicon sequence variant (ASV) feature tables. QIIME2/DADA2 were utilized for sequence visualization and quality trimming (Callahan et al., 2016). Taxonomy was aligned to the SILVA SSU (Ref NR release v138) reference database using the Naïve Bayes Classifier (Quast et al., 2012) Table 1 Pilot scale SSF operation and treatments applied for bioaugmentation. Pilot SSF number Metaldehyde in inlet Bioaugmentation Purpose Phase 1 (days 1–55) Purpose Phase 2 (days 56–72) 1 No No Non-treated control Non-treated control 2 No Yes Persistence of bioaugmentation agent A. calcoaceticus E1 without metaldehyde input Persistence of bioaugmentation agent A. calcoaceticus E1 without metaldehyde input 3 Yes Yes Effect of bioaugmentation with A. calcoaceticus E1 (1 × /2 × ) on metaldehyde removal – replicate 1 Effect of bioaugmentation with A. calcoaceticus E1 (3x) on metaldehyde removal 4 Yes Yes Effect of bioaugmentation with A. calcoaceticus E1 (1 × /2 × ) on metaldehyde removal – replicate 2 Effect of bioaugmentation with Sphingobium CMET-H on metaldehyde removal 5 Yes Yes Removal of metaldehyde without bioaugmentation – replicate 1 Effect of bioaugmentation with Sphingobium CMET-H on metaldehyde removal 6 Yes No Removal of metaldehyde without bioaugmentation – replicate 2 Removal of metaldehyde without bioaugmentation Fig. 1. Design and operation of the pilot scale SSF used for bioaugmentation experiments: A) pilot slow sand filter columns; B) covers used to suppress algal growth; C) C schematic of the pilot columns. L. Pickering et al. Chemosphere 363 (2024) 142956 5 and equal numbers of reads were rarefied to enable taxonomy analysis. Beta diversity was calculated using Bray-Curtis and UniFrac metrics, while Alpha diversity was determined based on Margalef’s (d) Richness index and Pielou’s evenness (J′). Where the Richness index is calculated d = (S - 1)/ln(N) where S is the number of species and N is the total number of individuals in the sample and Evenness is calculated based on J’ = H’/ln(S) where (H′) is the Shannon’s diversity index. Taxa assigned to the core microbiome were selected at a minimum prevalence of 0.9. 2.7. Quantification of total bacteria and known metaldehyde-degrading genes To assess the potential for metaldehyde degradation, the abundance of genes associated with different metaldehyde degradation pathways was quantified using qPCR. The mahY gene served as a marker for A. calcoaceticus E1, while mahS represented Sphingobium CMET-H. Both genes enable metaldehyde degradation and can be plasmid-derived, suggesting that their local increase could result from horizontal gene transfer (Castro-Gutiérrez et al., 2022a). Nonetheless, a previous study argued that this is unlikely to affect reactor performance due to the high concentration of bioaugmentation agents within the reactors. To eluci- date the relationship between augmented agents and aspects of the native microbiome, the total bacterial abundance was measured by qPCR using the 341F/534R primer pair (Petrić et al., 2011). Oligonu- cleotides used for degrader quantification are listed in Supplementary Table 1 qPCR conditions matched those used in earlier study (Cas- tro-Gutiérrez et al., 2022b). 2.8. Statistical analysis Statistical analysis of chemical performance data was conducted using SPSS (v25, IBM, USA), while microbial community data analysis was performed in PRIMER7 (Primer-E, Auckland, New Zealand). The 16S rRNA ASVs at various taxonomic levels were transformed using a square root function to balance the influence of abundant and less abundant taxa, allowing a more accurate assessment of sample simi- larities. The Bray-Curtis dissimilarity index, which measures the compositional dissimilarity between two different samples, was calcu- lated, and a resemblance matrix was constructed. PERMANOVA was employed to evaluate the influence of bioaugmentation, metaldehyde dosing, and filter level factors on microbial community composition (999 permutations). Principal coordinates analysis (PCoA) was used for data ordination. P-values were generated through PERMANOVA. One and two-way ANOVA and Tukey HSD post hoc tests were conducted at a 95% confidence level. Tukey HSD was suited to this data for identifying significant differences between group means while minimizing the risk of false positives. On the PCoA plot, Hierarchical Clustering Analysis (HCA) was employed to identify and visualize natural groupings within the dataset to identify relationships between samples and help draw conclusions about the underlying factors driving these patterns. BEST (BioEnv and Stepwise) analysis was used to identify biota which best matched observed patterns in community composition. 3. Results and discussion 3.1. Water quality and SSF performance during bioaugmentation Over the duration of the study, the influent water was between 0.7 and 1.1 NTU. The filter outlet turbidity ranged between 0.3 and 1.4 NTU. Direct comparison between the bioaugmented filters and controls (without bioaugmentation) showed that the bulk turbidity was variable 0.36 ± 0.44 (average ± standard deviation) between pilot scale filters, but the removal between these two groups of SSF did not differ in a significant way, suggesting the bioaugmentation did not compromise filtration efficiency for turbidity removal. Other water quality parame- ters such as pH ranged from 8.06 to 8.68 for the influent water and 7.78–9.64 for the filter outlets. The nitrate concentration was 0.3–1.3 mg L− 1 for filter effluent and 0.7–1.1 mg L− 1 for the influent. During the filters’ acclimation period (before bacteria dosing), there was limited removal of metaldehyde (<5.3% in all SSF with standard deviation± 4.5%). This fluctuation in the metaldehyde concentration in the filter outlet was attributed to acceptable variability in the experi- mental detection of the pesticide using the LCMS/MSmethod andminor variability expected at pilot scale. Chemical adsorption of metaldehyde is not an important removal pathway in sand media filters validated in earlier work (Rolph et al., 2018), but can influence reactor performance in the following ways. The sand or biofilm’s adsorptive properties significantly impact the availability of metaldehyde for microbial degradation. Enhanced adsorption reduces its presence in the aqueous phase, diminishing the effectiveness of bioaugmented bacteria. Conversely, metaldehyde that desorbs complicates removal control. Its moderate solubility and polarity allow interactions with both water and filter media, affecting the balance between adsorbed and dissolved states and thus the efficiency of microbial degradation. Filters 1 and 2 were not supplied with metaldehyde and served as experimental controls. In these cases, the pesticide was not detected in the influent or effluent of these SSF for the duration of the experiment. Phase 1 consisted of two bioaugmentation events (dose 1 and dose 2; Table 2). The first dose event consisted of a 1 × addition of A. calcoaceticus (2.4× 107 CFUmL− 1) in filter 3 and filter 4. Filters 5 and 6 were dosed with pesticide but were not dosed with bioaugmentation agents and no metaldehyde removal was observed during phase 1 (Table 2). This was despite>1 month of continuous metaldehyde dosing prior to this experimental monitoring programme. In this case, it was evident that bioenrichment of microbial degrading bacteria did not occur following spiking by metaldehyde. This was different to the ob- servations from Rolph et al. (2020) where pre-acclimated sand/biofilm obtained from a full scale SSF enriched at substantially higher metal- dehyde concentration (2 μg L− 1 compared to 50 μg L− 1) successfully achieved removal for a period of 100 days though not to compliance levels. This difference was explained by higher initial substrate con- centration and the use of pre-acclimated sand which was derived from a site already achieving biological metaldehyde removal. Angeles-de Paz et al. (2023) provides examples of successful bioaugmentation. The study showed that inoculating sewage sludge with Penicillium oxalicum and an enriched microbial consortium improved pharmaceutical com- pound degradation and reduced compost toxicity. Repeated inoculations Table 2 The metaldehyde removal performance of each filter during the two phases of operation. Bold indicates that compliance to <0.1 μg L− 1 pesticide concentration was reached. Period of operation Metaldehyde removed mean (min - max) F1 F2 F3 F4 F5 F6 Phase 1 Acclimation (n = 8) 0 0 5.4 (0–10.75) 3.4 (0–10.4) 3.2 (0–12.1) 4 (0–13) Phase 1 Dose event 1(n = 11) 0 0 9.5 (0–37.5) 11.6 (0–56.1) 0.4 (0–1.7) 3.2 (0–10.4) Phase 1 Dose event 2 (n = 9) 0 0 13.8 (0–74.6) 25.7 (0–78.1) 3.4 (0–12.3) 2.8 (0–0.11.2) Phase 2 Dose event 3 (n = 11) 0 0 12.8 (0–49.1) 57.3 (0–100) 75.3 (12.9–100) 3.3 (0–16.6) 0% removal is indicative of metaldehyde < LOD. L. Pickering et al. Chemosphere 363 (2024) 142956 6 under representative real conditions enhanced degradation performance compared to traditional methods, demonstrating the effectiveness of adapted microbial communities in improving product quality and environmental safety. For the bioaugmented filters the metaldehyde removal was low at 9.5%, which improved marginally to achieve a maximum of 37.5 % removal (Table 3). The replicate column (filter 4) had a similar average metaldehyde removal profile, with an average removal of 11.6% (range between <0.5 and 56.1%). After 5 days, the metaldehyde removal returned to pre-dose levels (2.2–4.7%) in both replicate filters, sug- gesting the bioaugmentation (A. calcoaceticus) agent did not establish within the SSF (either through adhesion, adsorption, or incorporation). It was hypothesized that poor removal was linked to insufficient abun- dance of the degrading strain or poor affinity for metabolising metal- dehyde in the up-scaled systems. Therefore, the next dose of A. calcoaceticus was increased to 2 × (4.84 × 107 CFU mL− 1) in both filters 3 and 4 (Tables 1and 2) in order to establish if a lack of metal- dehyde degraders was influencing the poor metaldehyde removal rates observed. However, the average metaldehyde removal was similar be- tween the 1 × and 2 × dosing regimens (13.8 and 25.7%), while the peak pesticide removal was similar at 74.6 and 78.1% for these filters, respectively. However, these filters did not reduce the pesticide to compliance levels of below 0.1 μg L, equivalent to a 95% reduction. The observed duration of metaldehyde removal was similar to that of a single dose, lasting 4–8 days, indicating potential loss of activity or displace- ment of the bacteria from the SSF which could explain transience of observed effect (Table 3). Similar results have been observed elsewhere using different strains and pesticides in pilot studies (Albers et al., 2015; Horemans et al., 2017; Sekhar et al., 2016; Hassard et al., 2022). Increased inoculum densities of Sphingobium CMET-H significantly improved both initial and ongoing metaldehyde degradation, under- scoring the importance of sufficient microbial populations for effective pollutant breakdown in sand filters. Effective management of inoculum density is crucial for enhancing bioaugmentation performance in water treatment processes. Higher initial doses of pollutant-degrading bacteria survived longer and were more effective in sequencing batch reactors, indicating that a larger microbial population can sustain bio- augmentation efforts (Chettri et al., 2024) and accelerates pollutant removal rates (Muter, 2023). During Phase 2, a 3× dose (8.11× 107 CFU mL− 1) of A. calcoaceticus was initiated to further explore whether the persistence of the pesticide removal effect could be improved. The 3 × dose showed lower perfor- mance (12.8%; 1.5–49.1%) than the 2 × dose, possibly due to compe- tition effects within biofilms (Miao et al., 2021). Previous bacteria tested therefore lacked the desired persistence, distribution, and activity within the SSF, prompting the trial of a strain of Sphingobium that per- formed well in an earlier study. Sphingobium is a Gram-negative, oxi- dase-positive, non-fermentative rod genera and our identified strain (Sphingobium CMET-H) has a plasmid-acquired metaldehyde degrada- tion pathway coded by the mahY gene, related to the vicinal oxygen chelate family (Castro-Gutiérrez et al., 2020). Sphingobium CMET-H strain (5 × 107 CFU mL− 1) was added to filters 4 and 5 to compare its sustained pesticide removal performance against A. calcoaceticus E1 (filter 5) and assess the influence of dosing both bioaugmentation agents on the SSF microbiome. Future studies should deploy additional methods, such as viability qPCR, could offer further insights into the survival and activity of these organisms. Samuelsen et al. (2017) found that Sphingomonadaceae family members, specifically Sphingomonas, exhibit favourable cell surface hydrophobicity and adherence capabil- ities. These features allow them to adhere to clean sand and establish in filter columns, and it is suggested here would promote sustained degradation of target pesticides even within real water matrix with established biofilm community in SSF. In filter 5, complete metaldehyde removal (<0.01 μg L− 1) occurred after one day following dosing and lasted for over 15 days, suggesting rapid establishment and activity. In filter 4, which had received a previous dose of A. calcoaceticus, the time to achieve complete removal of metaldehyde was slightly lower, taking 2 days. This may have been because of competition/inhibition effects. Future studies are needed to confirm the mechanisms behind the observed inhibition between augmented strains. 3.2. Impact of bioaugmentation on microbial community in SSF During the ripening phase of SSF, the microbial community un- dergoes colonization, diversification, and stabilization, leading to improved biofilm formation and contaminant removal efficiency. Thus, in the dynamic environment of SSF, understanding the ripening phase is essential for optimizing degrader dosing and formulating effective design strategies. Throughout the study, a change in the natural mi- crobial community was observed due to the filter ripening process across all SSF columns (PERMANOVA F (8, 44)= [4.1, p= 0.01]). The analysis of the control column, Filter 1, which was not subjected to an augmentation dose, revealed an interesting pattern of microbial change. Most notably, there was a significant increase from the beginning to the end of the trial in the relative abundance of the dominant taxa, which included members of the Proteobacteria and Planctomycetes groups, such as f_Ellin6075 (77.9% increase) and f_Isosphaeraceae (100% in- crease), along with f__Pirellulaceae; g__A17 (99% increase). However, a contrasting trend was seen with several other taxa. For instance, o_Myxococcales showed a stark decline of 1980%, while Rhodobacter and Acidovorax decreased by 994% and 4000% respectively. Methyl- otenera also diminished by 3348%. Other taxa, such as Reyranella and f__Cryomorphaceae, which are typically more consistent members of the SSF microbial community, showedmixed trends. Specifically, Reyranella showed a relative abundance decrease of about 20%, whereas f__Cryo- morphaceae’s relative abundance increased by approximately 45%. Observed shifts in taxa resulted in a shift towards a more diverse and cryptic microbial community (many microorganisms were taxonomi- cally unresolved) with time, a pattern also seen in other filters (2–6), suggesting that augmenting agents did not directly impact the filter ripening process (Fig. 2). This was consistent with earlier observations about bulk water quality determinants e.g., filtrate turbidity remaining broadly similar during filter run. It also shows the value of sufficient controls and where possible use of biological replicates during engi- neering biology experiments. Bioaugmentation with A. calcoaceticus E1 at either 1 × or 2 × con- centration (Filters 2–4) significantly affected the sand filter microbial community (PERMANOVA F (1, 75) = [36.9, p < 0.001], Fig. 2). This work was consistent with previous work stating that controlling (i.e., removing) for Acinetobacter sp. had significant impact on the Table 3 – Diversity statistics for the microbial community with filter depth. Total Species (S) Total Individuals (N) Margalef Species Richness (d) Pielou’s eveness (J′) Filter 1 Top 304 32058 29.20 0.63 Middle 306 30595 29.53 0.58 Lower 293 27647 28.55 0.61 Filter 2 Top 313 50586 28.8 0.56 Middle 318 57834 28.9 0.57 Lower 332 57564 30.2 0.58 Filter 3 Top 262 122904 22.27 0.13 Middle 315 52929 28.87 0.59 Lower 307 57189 27.93 0.60 Filter 4 Top 309 82186 27.22 0.39 Middle 385 102350 33.3 0.58 Lower 305 35536 29.012 0.62 Filter 5 Top 272 48474 25.12 0.56 Middle 289 46391 26.8 0.56 Lower 278 33281 26.6 0.57 Filter 6 Top 235 28414 22.82 0.63 Middle 308 49972 28.38 0.59 Lower 320 74002 28.45 0.57 L. Pickering et al. Chemosphere 363 (2024) 142956 7 background Schmutzdeckemicrobial community. This study showed that dosing Sphingobium CMET-H in filters 5 and 6 resulted minimal impacts to backgroundmicrobiome. During Phase 1, there was a local increase in the abundance of A. calcoaceticus E1, from 0.09 ± 0.004% of the total community before dosing to 75.2 ± 5.4% after the 1 × dose, based on Illumina sequence data. A. calcoaceticus E1 persisted in the surface layers of filters 2, 3, and 4, but its duration of 21 days after the first dosing in Phase 1 was insufficient to enable formation of a stable population for even a single SSF run (window between skim cleaning and period of steady-state operation). It was hypothesized that inadequate metaldehyde removal is related to the effective deactivation of pesticide-degrading enzymes within bacteria as they utilize other forms of assimilable organic carbon (AOC) for metabolism instead of the micropollutant. This phenomenon has occurred in mixed culture systems when pesticide concentrations are low or when enzyme affinities for the target pesticide are lower than those for other AOC substrates (Helbling, 2015). It is important to note that qPCR or sequencing methods for characterizing microbial com- munities can also detect genetic material from non-viable organisms, which could account for the persistence but poor performance of A. calcoaceticus E1 (Ho et al., 2020). However, given the evidence of A. calcoaceticus E1 strain aggregating into clusters, it may be expected that these cells would remain in the filter but exhibit poorer removal kinetics due to the large particle size of these bacterial assemblages, restricting mass transfer of pesticide to degrader. In contrast, with Sphingobium CMET-H, cell aggregation was not observed, and the rela- tive bacterial abundance was lower in the Schmutzdecke layers and it declined as the experiment progressed, suggesting lower accumulation and/or gradual washout of the organism from the filter (Fig. 2). A constant flow of water through the filter does not mean that cells will leave the reactor due to the adhesion of cells to the media and potential incorporation in biofilms assuming positive selection. Importantly, this coincided with effective metaldehyde removal, indicating that small populations of established degraders are sufficient for degradation of micropollutants. Filter 4 received a dose of A. calcoaceticus E1 in phase 1 and Sphin- gobium CMET-H in phase 2. The A. calcoaceticus E1 persisted at an elevated abundance on the SSF surface (32.9%) compared to the reads associated with Sphingobium CMET-H (reads from order Sphingomona- dales) (5.3%). This effect was observed immediately after dosing and continued until the end of the study, with A. calcoaceticus E1 repre- senting 42.9% of the Schmutzdecke abundance while Sphingobium CMET- H was not detected. In filter 5, the order Sphingomonadales was present at trace levels before Phase 2 dosing. However, it increased its relative abundance by 99% after dosing, suggesting that the increase was due to the bioaugmentation agent and that this was different to the 53% in- crease in filter 4 which was pre-dosed with the other bioaugmentation agent first – suggesting inhibition when predosed with another organ- ism. The 16s analysis produced data which was specific enough to identify the genus level of Sphingobium but could not further resolve to strains that were positive for the metaldehyde degrading plasmids directly. However, comparison between the Illumina data with the qPCR data for degrading gene abundance was used to resolve differences observed. In summary, impacts of Acinetobacter calcoaceticus E1 and Sphingobium CMET-H on filter performance. Acinetobacter calcoaceticus E1 proved more effective in the top layer due to its aggregation and surface hydrophobicity, promoting adhesion but limiting deeper pene- tration. This results in high initial activity but reduced long-term per- formance. Sphingobium CMET-H, however, exhibited better distribution throughout the filter, due to its superior cell surface properties and plasmid-encoded degradation pathways, maintaining high efficiency even at lower densities. Sequential dosing showed that introducing Acinetobacter calcoaceticus E1 first hindered the establishment and per- formance of Sphingobium CMET-H, whereas the reverse sequence allowed more uniform colonization and consistent degradation. It was hypothesized that the most significant effect of bio- augmentation would occur near the dosing point, as increased compe- tition between the native microbiome and the augmented strains for resources and niche spaces can influence the persistence of augmented strains in reactors. To investigate this, a Bray-Curtis analysis of the bacterial communities at different depths, revealing important differ- ences within and between the filters. For instance, there was a signifi- cant difference between the bacterial communities in the Schmutzdecke (top) and the subsurface layers of the sand filter (PERMANOVA level F Fig. 2. Background microbial community in the top layer of the slow sand filter columns (Schmutzdecke), showing a genus level relative abundance of 16S rRNA gene copies through the duration of the study (restricted to the top 20 genera or next available taxonomic classification). The next available taxonomic classifications is defined as considering: o = order, f = family, k = kingdom where genera was not available based on comparison to database. PERMANOVA statistics to examine the effects on the variability of the microbiome between different filters, where Bioaug = bioaugmentation; Sphingo = Sphingobium sp. CMET-H; Acineto = Acineto- bacteria E1. Each sample event is shown and displayed as vertical stacked bar. The sampling events were on days 1, 16, 18, 23, 30, 37, 42, 44, 51, 56, 58, 65, and 72. L. Pickering et al. Chemosphere 363 (2024) 142956 8 (2, 10)= [11.8, p< 0.001]), with between filter effects also contributing to this variability (p< 0.001, Fig. 3A). The taxa most responsible for this difference in each layer at 90% threshold (arbitrary based on HCA) in the top layer of filters 3 and 4 was Acinetobacter. In contrast in the other top layers (filters 1,2,5 and 6) the taxa Defluviimonas and organism SWB02 most contributed to observed variability (Supplementary Fig. 1). In contrast, for the rest of the filters and their respective layers at 90% HCA threshold, the organisms MND1, Lacunisphaera and Fluviicola were most important components governing the microbiome variability. A diverse group of taxa including Arenimonas was poorly linked (50% - HCA) and only associated strongly with the middle and lower layers of filters dosed with metaldehyde but not A. calcoaceticus E1 suggesting they might be commensals either growing on metaldehyde or benefit- ting from its application. Other studies help to elucidate what might be influencing this as they have identified that physical bacteria removal occurs primarily in the top section of ripened SSF beds due to prefer- ential particle accumulation. Yet, in systems that are either poorly ripened or recently skimmed, this process is less effective. This in- efficiency is attributed to variations in the pore size radius, which are related to the presence or absence of biofilm formation (Trikannad et al., 2023). The co-accumulation and filtering of both organic material and microbiota presumably contribute to the observed stratification within the SSF biofilm with different bed depths (Campos et al., 2002). In the PCoA (Fig. 3B), the PCO1 axis distinguishes variability between the top and middle- and lower-layer communities, while PCO2 shows the in- fluence of A. calcoaceticus E1 dosing on the Schmutzdecke microbial community. In the lower layers, there was a significant difference in the filter’s microbial taxa, with PCO1, differentiating the beta diversity of filters augmented with A. calcoaceticus E1 compared to those without. The dosing of A. calcoaceticus E1 caused a pronounced shift in community diversity, particularly in the top layers of the sand filters (PERMANOVA pairwise test for levels, p < 0.001, Fig. 3A). However, alpha diversity measures (within sample comparisons) remained mostly unaffected by either pesticide dosing or bioaugmentation agents (Table 3). For instance, Pielou’s evenness and richness values did not change substantially. The exception was the microbial community in the top layer of filter 3, where the Margalef (d) Richness value (dimen- sionless) was 22.27 compared to 28.55± 1.63 for the remaining samples (Table 3). In filter 6, the d value was also low at 22.82. The decrease in d value in filter 3, along with an evenness value of 0.13 (below the average of 0.56 ± 0.11 for other filter communities), suggested an SSF biofilm community that was dominated by the bioaugmentation agent at the end of the trial (Table 3). A 6 point change in the Margalef richness is considered a significant change, indicating a notable shift in the mi- crobial community composition and potentially reflecting the impact of bioaugmentation. However, further analysis is required to determine the Fig. 3. A) Principal coordinate analysis (PCoA) plot of weighted UniFrac distance. Principal coordinate analysis was used to plot the beta diversity of the genus level community for different filter levels. Samples were obtained following destructive sampling of the filter bed at the end of the study; B) PCoA plot containing only the community data from the lower and middle levels of the filter; C) Heat map showing relative abundance of microorganisms at different levels of the filter, based on genus level classification – samples were taken at the end of the study. Analysis was restricted to the top 20 taxa. L. Pickering et al. Chemosphere 363 (2024) 142956 9 ecological and functional relevance of this change. The data analysis presented indicates that the systematic and continuous dosing of met- aldehyde during the 72-day study and prior acclimation periods influ- enced both the surface and subsurface bacterial communities in the filter (Fig. 3A and B). However, limited differences were observed when comparing the alpha diversity scores for the different samples (Table 3). This could suggest that while the structure of bacterial communities was affected by metaldehyde exposure, the overall diversity within these communities remained relatively stable, possibly due to a resilience mechanism or the presence of metaldehyde at micropollutant levels exerting minimal impact to the microbial population. To explore which taxa are contributing to the observed differences a heatmap was plotted (Fig. 3C). Filter 1 did not have bioaugmentation or metaldehyde dosing and had a distinct microbial community which was differentiated on the PCO2 axis, with communities being 73% similar based on HCA. The bioaugmentation of Sphingobium CMET-H had a smaller impact on microbiome of in the lower layers compared to the middle subsurface layers due to similar clustering between these co- ordinates and filter biofilm not subject to any augmentation (Fig. 3B). Analysis of the genera within filters and across different layers revealed that A. calcoaceticus E1 persisted in the Schmutzdecke 30 days after the last dosing event. In filter 5, which was bioaugmented with Sphingobium CMET-H, a higher percentage of reads associated with Sphingopyxis and Sphinorhabdus (order Sphingomonadales) was found compared to other filters. In the middle and lower layers, Arenimonas accounted for 17.1% and 18.4% of the taxa, respectively, but was absent from the top layer. In contrast, other filters had lower abundance of this taxon (around 5% for middle layers and between 0 and >4.1% for lower layers; Fig. 3C). This is noteworthy since some Arenimonas species are known for their bioremediation potential, degrading and utilizing environmental pol- lutants for carbon and energy sources (Wang et al., 2020). Their meta- bolic capabilities enable them to play a role in nutrient cycling and ecosystem balance (Makk et al., 2015). BEST analysis which looks for correlations amongst biota revealed a 95% correlation between Areni- monas and MND1, Lacunisphaera, Sericytochromatia, Fim- briimonadaceae. This is indicative of a polybacterial shift in the microbiome associated with bioaugmentation. However, the effect of bioaugmentation on the presence and ecological roles of Arenimonas and related taxa requires further investigation. Future research should aim to understand the diversity and distribution of these organisms and determine if their co-existence with Sphingobium CMET-H boosts pesti- cide degradation and the ecological effectiveness of bioaugmentation in reactors. It will also explore whether this association is synergistic, coincidental, or influenced by factors like the carrier solutions used to dose the bioaugmentation agent on subsequent performance. Moreover, the absence of Arenimonas from the top layer in all but the control SSF column suggests it occupies a niche within the SSF microbiome, possibly at lower depths where there are reduced toxic effects from pesticides or less competition with either native or augmented strains. 3.3. Activity of bioaugmentation agents At day 51 before the first dosing event with Sphingobium CMET-H, there was 7.56 × 107 ASVs and 3.45 × 107 ASVs of Sphingobium sp. in filters 4 and 5, respectively. These represented Sphingobium strains that were resident in the background microbiome. These organisms were not removing metaldehyde, suggesting that these strains of Sphingobium did not have the capability to degrade metaldehyde, or were inactive. On day 56, Sphingobium CMET-H was dosed into the filter and this had the effect of increasing the relative abundance to 2.24× 1012 ASVs and 5.67 × 1012 ASVs in filters 4 and 5. As the experiment progressed there was a steady decline in the abundance of Sphingobium CMET-H. After 16 days following the bioaugmentation dosing, the reactor abundance decreased to 5.4 × 109 ASVs and 3.33 × 1010 ASVs in filters 4 and 5. The specific removal rate increased gradually from 4.8 pg.metaldehyde.mahS− 1. day− 1 at day 58 to 1.52× 102 pg.metaldehyde.mahS− 1.day− 1 suggesting that the metaldehyde degraders had a greater specific activity. Filter 5 had a lower specific removal rate than filter 4 at all-time points but attained removal of metaldehyde quicker (Table 2). At day 72 the gap between the removal rates were most pronounced, with a 5-fold dif- ference between the specific removal of metaldehyde in filters 4 and 5, driven in part by a lower degrader cell count in filter 4 for equivalent metaldehyde removal (Fig. 4A). This was further evidence that the previous dosing of A. calcoaceticus E1 had a negative impact on the establishment and persistence of Sphingobium sp. CMET-H in the filters. The relationship between microbial activity and abundance within the filters was further investigated across different depths. This analysis of microbial abundance can be framed around conventional SSF theory proposed by Huisman and Wood (1974), which suggests a higher mi- crobial abundance occurs towards the top layer of the filters due to alga growth, accumulation of bacteria and particles, and elevated bacterial growth rates. Consistent with this theory, the top layer exhibited a significantly higher microbial abundance, registering 15.1 × 106 ± 6.0 × 104 and 2.3 × 106 ± 1.22 × 105 16s rRNA gene copies in filters 4 and 5, respectively. This abundance data, when interpreted in context, helps to affirm the conventional SSF theory and helps shape understanding of microbial dynamics within these systems. To this end, the Prokaryotic abundance saw a gradual reduction from the top to the lower layers (Fig. 4B). To understand the distribution of Sphingobium sp. within the bed, we calculated the percentage of mahS genes relative to the total number of 16s rRNA genes determined by PCR. For filter 4, mahS gene copies were found to be 0.054%, 0.28%, and 0.36% in the top, middle, and lower layers, respectively. In contrast, these values were 0.26%, 2.8%, and 2.2% for filter 5 (Fig. 3C). Despite the diversity and abun- dance of the background microbial community, by the end of the experiment, Sphingobium sp. CMET-H was primarily located in the filter bed’s top layer (53% and 60% in filters 4 and 5, respectively). The Sphingobium sp. exhibited better distribution throughout the bed, most notably in filter 5, where the metaldehyde-degrading population esca- lated from 0.26% at the surface to 2.8% in the middle layer, before decreasing slightly to 2.2% in the lower strata. This indicates the met- aldehyde degrader’s ability to disperse and form a small part of the community, an effect less pronounced in filter 4, which had previously been bioaugmented with a different agent. 3.4. Why do some bioagumentation agents perform better than others? This study underscores the important role of bioaugmentation, spe- cifically the introduction of pesticide-degrading bacteria, in enhancing the performance of SSF in removing recalcitrant pesticides from drink- ing water. The research particularly highlights the efficacy of the bac- terial strain, Sphingobium CMET-H, which demonstrated consistent (near 100%) metaldehyde removal over a 15-day period. Furthermore, this study reveals the significant influence of the dosing sequence of bio- augmented bacteria on the functional pesticide removal capacity of the filter. It highlights that the improper dosing sequence can impede the establishment of effective bacteria strains, consequently reducing the filter’s ability to degrade pesticides. This research paves the way for a well-defined approach to establishing resilient pesticide remediation in drinking water treatment. It advocates for a strategy that achieves effective micropollutant degradation and also preserves the integrity of the filter microbiome, an aspect that is of paramount importance considering the emerging threats posed by new micropollutants to drinking water supplies. Elevated dosing of bioaugmentation agents in biofilms might lead to competition for limited resources, changing microbial dynamics, potentially inhibiting contaminant degradation. Aggregation into clumps, driven by cell surface characteristics, extracellular substances, and fluid dynamics, can reduce the effectiveness of these agents. Addressing water chemistry through surfactants or cell wall modifica- tions could prevent such aggregation, ensuring better contact with pollutants. Bioaugmentation failures have been linked to predation L. Pickering et al. Chemosphere 363 (2024) 142956 10 (Ellegaard-Jensen et al., 2016), washout (Sørensen et al., 2007), and genetic factors like plasmid loss (Rios Miguel et al., 2020; Vargas and Hattori, 1986), complicating the maintenance of active degraders. Reducing flow rates can enhance microbial contact with pollutants, improving degradation (Horemans et al., 2017). Optimizing inoculation strategies, including periodic reintroductions, might leverage ecological principles to support degrader establishment. This approach would utilize times of available resources and less native competition, helping Fig. 4. A) Specific metaldehyde removal rate at different time points during phase 2 (days 56–72 of study). Sequence data expressed as the total number of Sphingobium specific reads B) prokaryotic abundance based on qPCR analysis of 16s rRNA gene amplicons; C) percentage of SSF microbial community comprised of metaldehyde degraders based on analysis of 16s rRNA and mahS genes using qPCR. L. Pickering et al. Chemosphere 363 (2024) 142956 11 introduced microbes to integrate and endure within the microbial community. A successful bioaugmentation strategy aims for the wide- spread establishment of degraders, providing resilience against disrup- tions and ensuring sustained pesticide degradation (Hassard et al., 2022). The optimal time to dose bioaugmentation agents during an SSF ripening cycle is typically after the initial biofilm has established and stabilized, which usually occurs around 2–3 months into the ripening phase. This timing ensures that the native microbial community is robust enough to support and interact beneficially with the introduced strains, maximizing the efficacy of the bioaugmentation process. An alternative hypothesis is to dose organisms earlier to accelerate ripening (Rosenqvist et al., 2024) and gain functional benefits, such as enhanced pesticide removal, as studied here. This research is considered to have the following impacts, firstly on sustainable water management practices, specifically in the context of biofiltration for drinking water treatment. The successful use of the bacterial strain Sphingobium CMET-H in degrading persistent pesticides like metaldehyde can inspire similar bioaugmentation strategies for a wide range of contaminants. By elucidating the importance of dosing sequences, this research can inform improved operational protocols to maximize the efficiency of bioaugmentation in SSF and possible be translated to other relevant granular media systems such as RGF and GAC systems. The findings can also contribute to the development of more sustainable pesticide management practices. For instance, effec- tive pesticide degradation in drinking water treatment can reduce the need for chemical disinfectants, leading to less chemical waste and lower energy consumption. In turn, this can lower/mitigate the envi- ronmental impact of water treatment processes. Moreover, the demon- strated potential of bioaugmentation for preserving the filter microbiome can promote biodiversity within these systems. This could enhance the resilience of SSF to disruptions, leading to more reliable and sustainable water treatment outcomes. This research could also be applied to wastewater treatment. The removal of pesticides and other micropollutants from wastewater is a major challenge, and the suc- cessful bioaugmentation strategy demonstrated here could be adapted for use in biofilters or constructed wetlands for wastewater treatment. This research offers valuable insights for more sustainable water and waste management practices. It highlights the potential of bio- augmentation for pesticide treatment in run-off interception technolo- gies and soil remediation, offering low-energy, biologically-driven solutions that are climate-resilient in that they do not contribute further to the issue. The findings could also inform the development of a flexible toolkit for introducing smart dosing approaches linked to online catch- ment sensors whereby feedforward/feedback predictive control systems and dosing algorithms could apply the correct dosage of specific de- graders into water treatment systems to meet a variety of transient and evolving pesticide threats. The demonstrated effectiveness of bio- augmentation suggests its potential as a ‘chassis’ or platform for addressing new and emerging pesticides, contributing to more robust water quality management. Future work could focus on optimizing system conditions to enhance the persistence and survival of introduced degraders and developing a broader range of pesticide-degrading bac- teria. Taken as a whole, this research paves the way for innovative, sustainable, and adaptable strategies for managing pesticide pollution in water and soil. 3.5. Health implications of biological pesticide degradation The findings from this study have significant health implications for drinking water safety. By demonstrating that bioaugmentation with Sphingobium CMET-H can consistently reduce metaldehyde concen- trations to below regulatory limits, this research offers a viable strategy for mitigating the health risks of pesticide contamination. Pesticides like metaldehyde pose serious health risks due to their potential toxicity and persistence in the environment. Chronic exposure, even at low levels, can lead to endocrine disruption, reproductive issues, increased cancer risk and changes to the gut microbiome and its interaction with health (Matsuzaki et al., 2023). Ensuring drinking water is free from these substances is critical for public health. The study also highlights the importance of microbiome stability in water treatment systems. A well-functioning microbial community enhances contaminant degrada- tion and prevents the proliferation of pathogenic bacteria, further safeguarding water quality and health. Metaldehyde breaks down into several intermediate compounds during the degradation process, including acetaldehyde, formaldehyde, and acetic acid. The toxicity of these metabolites varies. Acetaldehyde and formaldehyde are known to be toxic; acetaldehyde is a probable human carcinogen, while formaldehyde is classified as a known human carcinogen. Both compounds can have adverse health effects even at low concentrations, potentially affecting the quality of drinking water. The study did not specifically measure the concentrations of these interme- diate substances in the water following the degradation of metaldehyde. However, their presence can significantly impact water quality and pose health risks. Future research should focus on identifying and quantifying these breakdown products to ensure that the degradation process does not inadvertently compromise water safety. Additionally, understanding the formation and fate of these intermediates will help in designing more effective water treatment systems that not only degrade the primary contaminant but also mitigate the risks associated with its byproducts. 3.6. Application of bioaugmentation in drinking water treatment The findings of this study offer several promising industrial appli- cations, particularly in the field of water treatment. The successful bioaugmentation with Sphingobium CMET-H to reduce metaldehyde levels below regulatory limits can be scaled up and applied in large-scale water treatment facilities. This method provides an efficient and cost- effective alternative to conventional chemical treatments, potentially reducing operational costs and environmental impact. Moreover, the insights gained into the interaction between bioaugmented strains and native microbial communities can be leveraged to optimize bioreactors and sand filters used in various industries. This can enhance the degradation of other persistent pollutants, including industrial chem- icals and pharmaceutical residues, making water reclamation processes more effective. The approach can also be adapted for use in agricultural runoff treatment systems to mitigate pesticide contamination at the source, protecting both surface and groundwater quality. Additionally, industries involved in soil remediation can apply these bioaugmentation strategies to degrade persistent pollutants, improving soil health and reducing the risk of environmental contamination. Furthermore, synthetic biology could play a crucial role in engi- neering more efficient bioaugmentation strains, tailored to target a wider range of contaminants with greater efficacy and resilience. Either of the strains tested in this study, Sphingobium CMET-H or Acinetobacter calcoaceticus E1, could act as a chassis organism for developing enhanced biodegradation capabilities. This work could also serve as a template to enhance biological treatments for more recalcitrant micro- pollutants, such as per- and polyfluoroalkyl substances (PFAS), thereby addressing a broader spectrum of environmental contaminants (Hassard et al., 2024). Overall, the study’s outcomes can drive innovations in sustainable water and soil management practices across various sectors, contributing to improved environmental and public health. The manuscript suggests several avenues for future work to enhance the pesticide removal efficiency of biological drinking water filters. These include developing and testing new bioaugmentation strains with broader or more potent degradation capabilities, as well as exploring the potential of combining multiple strains to create synergistic effects. Investigating the impact of varying operational parameters, such as flow rates and nutrient availability, on bioaugmentation efficacy could also yield valuable insights. In vitro studies are needed to test whether Aci- netobacter calcoaceticus E1 suppresses the survival of Sphingobium CMET- H through resource/space competition, or if it dominates in number, or L. Pickering et al. Chemosphere 363 (2024) 142956 12 it exhibits an antagonistic activity. Additionally, integrating advanced monitoring techniques, such as real-time biosensors, to track microbial community dynamics and contaminant levels could improve the man- agement and optimization of these systems. The study’s limitations include the relatively short experimental duration and the need for more comprehensive field trials to validate the findings under diverse envi- ronmental conditions. 4. Conclusion This study found that bioaugmentation in a continuous sand filtra- tion system effectively reduced metaldehyde in real water to below drinking water standards (0.1 μg L-1). The rate of removal varied by bioaugmentation agent, with Sphingobium CMET-H outperforming A. calcoaceticus E1 in both efficiency and duration of removal (>15 days vs. <5 days). The presence of both agents significantly altered the bio- film community structure. A. calcoaceticus E1 had a more pronounced effect on the top layer microbiome due to higher aggregation, leading to decreased diversity there. In contrast, Sphingobium CMET-H treatment resulted in a more evenly distributed and diverse microbial community, potentially due to its efficient metabolism and less competitive inter- action with native microbes. The study also highlighted possible syn- ergistic relationships between certain taxa like Arenimonas and the pesticide-degrading Sphingobium CMET-H, suggesting deeper niche occupation away from surface toxicity or microbial competition. How- ever, further research is necessary to understand these microbial dy- namics fully. Sphingobium CMET-H’s more uniform distribution within the filter facilitated consistent and extended metaldehyde degradation, underscoring the importance of microbial distribution, its interaction with contact time for biodegradation and therefore filter performance. This points to the effectiveness of Sphingobium CMET-H for prolonged pesticide degradation, supported by its resilience and positive in- teractions within the filter environment. The study suggests a viable approach for managing challenging pesticides and calls for further investigation into sustainable microbial strategies against novel and persistent organic micropollutants. Data access Data associated with this manuscript is available at: 10.17862/ cranfield.rd.21029239. CRediT authorship contribution statement Laura Pickering: Writing – original draft, Visualization, Methodol- ogy, Formal analysis, Data curation. Victor Castro-Gutierrez:Writing – original draft, Methodology, Formal analysis, Data curation, Conceptu- alization. Barrie Holden:Writing – original draft, Supervision, Funding acquisition, Conceptualization. John Haley:Writing – review& editing, Funding acquisition. Peter Jarvis: Writing – review & editing, Super- vision, Methodology, Investigation. Pablo Campo: Writing – review & editing, Supervision, Methodology, Funding acquisition. Francis Has- sard: Writing – review & editing, Supervision, Project administration, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. The decision to publish the research rested solely with the authors and the funder did not influence the de- cision to publish the research. Data availability Data will be made available on request. Acknowledgments The authors gratefully acknowledge financial support from the En- gineering and Physical Sciences Research Council (ESPRC) through their funding of a Doctoral Training Allocation Award to LP (EP/R513027/1) and from the project sponsors UK Water Industry Research (UKWIR Ltd.). The authors acknowledge Affinity Water for their assistance dur- ing the work. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.chemosphere.2024.142956. References Albers, C.N., Feld, L., Ellegaard-Jensen, L., Aamand, J., 2015. Degradation of trace concentrations of the persistent groundwater pollutant 2,6-dichlorobenzamide (BAM) in bioaugmented rapid sand filters. Water Res. 83, 61–70. https://doi.org/ 10.1016/j.watres.2015.06.023. Albers, C.N., Jacobsen, O.S., Aamand, J., 2014. Using 2,6-dichlorobenzamide (BAM) degrading Aminobacter sp. MSH1 in flow through biofilters - initial adhesion and BAM degradation potentials. Appl. Microbiol. Biotechnol. 98, 957–967. https://doi. org/10.1007/s00253-013-4942-6. Aldas-Vargas, A., van der Vooren, T., Rijnaarts, H.H.M., Sutton, N.B., 2021. Biostimulation is a valuable tool to assess pesticide biodegradation capacity of groundwater microorganisms. Chemosphere 280. https://doi.org/10.1016/j. chemosphere.2021.130793. Angeles-de Paz, G., León-Morcillo, R., Guzmán, S., Robledo-Mahón, T., Pozo, C., Calvo, C., Aranda, E., 2023. Pharmaceutical active compounds in sewage sludge: degradation improvement and conversion into an organic amendment by bioaugmentation-composting processes. Waste Manag. 168, 167–178. Balashova, N., Hiscock, K.M., Reid, B.J., Reynolds, R., 2021. Trends in metaldehyde concentrations and fluxes in a lowland, semi-agricultural catchment in the UK (2008–2018). Sci. Total Environ. 795 https://doi.org/10.1016/j. scitotenv.2021.148858. Bouchez, T., Patureau, D., Dabert, P., Juretschko, S., Doré, J., Delgenès, P., Moletta, R., Wagner, M., 2000. Ecological study of a bioaugmentation failure. Environ. Microbiol. 2, 179–190. https://doi.org/10.1046/j.1462-2920.2000.00091.x. Castro-Gutiérrez, V., Fuller, E., Thomas, J.C., Sinclair, C.J., Johnson, S., Helgason, T., Moir, J.W.B., 2020. Genomic basis for pesticide degradation revealed by selection, isolation and characterisation of a library of metaldehyde-degrading strains from soil. Soil Biol. Biochem. 142 https://doi.org/10.1016/j.soilbio.2019.107702. Castro-Gutiérrez, V., Hassard, F., Moir, J., 2022a. Probe-based qPCR assay enables the rapid and specific detection of bacterial degrading genes for the pesticide metaldehyde in soil. J. Microbiol. Methods 195. https://doi.org/10.1016/j. mimet.2022.106447. Castro-Gutiérrez, V., Pickering, L., Cambronero-Heinrichs, J., Holden, B., Haley, J., Jarvis, P., Jefferson, B., Helgason, T., Moir, J., Hassard, F., 2022b. Bioaugmentation of pilot-scale slow sand filters can achieve compliant levels for the micropollutant metaldehyde in a real water matrix. Water Res. 211, 118071 https://doi.org/ 10.1016/j.watres.2022.118071. Chettri, D., Verma, A.K., Verma, A.K., 2024. Bioaugmentation: an approach to biological treatment of pollutants. Biodegradation 35 (2), 117–135. Cosgrove, S., Jefferson, B., Jarvis, P., 2022. Application of activated carbon fabric for the removal of a recalcitrant pesticide from agricultural run-off. Sci. Total Environ. 815 https://doi.org/10.1016/j.scitotenv.2021.152626. Council Directive 98/83/EC, 2010. The European union drinking water directive. Documents in European Community Environmental Law. https://doi.org/10.1017/ cbo9780511610851.055. Dutta, N., Usman, M., Ashraf, M.A., Luo, G., Zhang, S., 2022. A critical review of recent advances in the bio-remediation of chlorinated substances by microbial dechlorinators. Chemical Engineering Journal Advances. https://doi.org/10.1016/j. ceja.2022.100359. Ellegaard-Jensen, L., Albers, C.N., Aamand, J., 2016. Protozoa graze on the 2,6- dichlorobenzamide (BAM)-degrading bacterium Aminobacter sp. MSH1 introduced into waterworks sand filters. Appl. Microbiol. Biotechnol. 100, 8965–8973. https:// doi.org/10.1007/s00253-016-7710-6. Fenner, K., Canonica, S., Wackett, L.P., Elsner, M., 2013. Evaluating pesticide degradation in the environment: blind spots and emerging opportunities. Science 341 (6147), 752–758. Gilliom, R.J., Barbash, J.E., Crawford, C.G., Hamilton, P.A., Martin, J.D., Nakagaki, N., Nowell, L.H., Scott, J.C., Stackelberg, P.E., Thelin, G.P., Wolock, D.M., 1992. circ1291.pdf - the Quality of Our Nation’s Waters—Pesticides in the Nation’s Streams and Ground Water, pp. 1992–2001. L. Pickering et al. https://doi.org/10.1016/j.chemosphere.2024.142956 https://doi.org/10.1016/j.chemosphere.2024.142956 https://doi.org/10.1016/j.watres.2015.06.023 https://doi.org/10.1016/j.watres.2015.06.023 https://doi.org/10.1007/s00253-013-4942-6 https://doi.org/10.1007/s00253-013-4942-6 https://doi.org/10.1016/j.chemosphere.2021.130793 https://doi.org/10.1016/j.chemosphere.2021.130793 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref4 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref4 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref4 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref4 https://doi.org/10.1016/j.scitotenv.2021.148858 https://doi.org/10.1016/j.scitotenv.2021.148858 https://doi.org/10.1046/j.1462-2920.2000.00091.x https://doi.org/10.1016/j.soilbio.2019.107702 https://doi.org/10.1016/j.mimet.2022.106447 https://doi.org/10.1016/j.mimet.2022.106447 https://doi.org/10.1016/j.watres.2022.118071 https://doi.org/10.1016/j.watres.2022.118071 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref11 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref11 https://doi.org/10.1016/j.scitotenv.2021.152626 https://doi.org/10.1017/cbo9780511610851.055 https://doi.org/10.1017/cbo9780511610851.055 https://doi.org/10.1016/j.ceja.2022.100359 https://doi.org/10.1016/j.ceja.2022.100359 https://doi.org/10.1007/s00253-016-7710-6 https://doi.org/10.1007/s00253-016-7710-6 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref16 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref16 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref16 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref17 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref17 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref17 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref17 Chemosphere 363 (2024) 142956 13 Haig, S.J., Schirmer, M., D’Amore, R., Gibbs, J., Davies, R.L., Collins, G., Quince, C., 2015. Stable-isotope probing and metagenomics reveal predation by protozoa drives E. Coli removal in slow sand filters. ISME J. 9, 797–808. https://doi.org/10.1038/ ismej.2014.175. Hassard, F., Elemo, T., Chipps, M., Turner, A., Jefferson, B., Graham, N., 2022. Underwater remote skimming of slow sand filters for sustainable water production. ACS ES and T Water. https://doi.org/10.1021/acsestwater.2c00313. Hassard, F., Curtis, T.P., Dotro, G.C., Golyshin, P., Gutierrez, T., Heaven, S., Horsfall, L., Jefferson, B., Jones, D.L., Krasnogor, N., Kumar, V., 2024. Scaling-up engineering biology for enhanced environmental solutions. ACS Synth. Biol. 13 (6), 1586–1588. Hassard, F., Gwyther, C.L., Farkas, K., Andrews, A., Jones, V., Cox, B., Brett, H., Jones, D. L., McDonald, J.E., Malham, S.K., 2016. Abundance and distribution of enteric bacteria and viruses in coastal and estuarine sediments-A review. Front. Microbiol. https://doi.org/10.3389/fmicb.2016.01692. Horemans, B., Raes, B., Vandermaesen, J., Simanjuntak, Y., Brocatus, H., T’Syen, J., Degryse, J., Boonen, J., Wittebol, J., Lapanje, A., Sørensen, S.R., Springael, D., 2017. Biocarriers improve bioaugmentation efficiency of a rapid sand filter for the treatment of 2,6-dichlorobenzamide-contaminated drinking water. Environ. Sci. Technol. 51, 1616–1625. https://doi.org/10.1021/acs.est.6b05027. Huisman, L., Wood, W.E., 1974. Slow Sand Filtration. World Health Organisation (WHO), Geneva. Geneva. Li, Y., Su, J., Ali, A., Hao, Z., Li, M., Yang, W., Wang, Z., 2022. Simultaneous removal of nitrate and heavy metals in a biofilm reactor filled with modified biochar. Sci. Total Environ. 851, 158175 https://doi.org/10.1016/j.scitotenv.2022.158175. Ma, H., Zhao, Y., Yang, K., Wang, Y., Zhang, C., Ji, M., 2022. Application oriented bioaugmentation processes: mechanism, performance improvement and scale-up. Bioresour. Technol. https://doi.org/10.1016/j.biortech.2021.126192. Makk, J., Homonnay, Z.G., Keki, Z., Nemes-Barnas, K., Marialigeti, K., Schumann, P., Toth, E.M., 2015. Arenimonas subflava sp. nov., isolated from a drinking water network, and emended description of the genus Arenimonas. Int. J. Syst. Evol. Microbiol. 65 (Pt_6), 1915–1921. Matsuzaki, R., Gunnigle, E., Geissen, V., Clarke, G., Nagpal, J., Cryan, J.F., 2023. Pesticide exposure and the microbiota-gut-brain axis. ISME J. 17 (8), 1153–1166. McDowall, B., Hoefel, D., Newcombe, G., Saint, C.P., Ho, L., 2009. Enhancing the biofiltration of geosmin by seeding sand filter columns with a consortium of geosmin-degrading bacteria. Water Res. 43, 433–440. https://doi.org/10.1016/j. watres.2008.10.044. Miao, Y., Heintz, M.B., Bell, C.H., Johnson, N.W., Polasko, A.L.P., Favero, D., Mahendra, S., 2021. Profiling microbial community structures and functions in bioremediation strategies for treating 1,4-dioxane-contaminated groundwater. J. Hazard Mater. 408 https://doi.org/10.1016/j.jhazmat.2020.124457. Mohamad Ibrahim, I.H., Gilfoyle, L., Reynolds, R., Voulvoulis, N., 2019. Integrated catchment management for reducing pesticide levels in water: engaging with stakeholders in East Anglia to tackle metaldehyde. Sci. Total Environ. 656, 1436–1447. https://doi.org/10.1016/j.scitotenv.2018.11.260. Muter, O., 2023. Current trends in bioaugmentation tools for bioremediation: a critical review of advances and knowledge gaps. Microorganisms 11 (3), 710. Petrić, I., Bru, D., Udiković-Kolić, N., Hršak, D., Philippot, L., Martin-Laurent, F., 2011. Evidence for shifts in the structure and abundance of the microbial community in a long-term PCB-contaminated soil under bioremediation. J. Hazard Mater. 195, 254–260. https://doi.org/10.1016/j.jhazmat.2011.08.036. Pinilla-Redondo, R., Olesen, A.K., Russel, J., de Vries, L.E., Christensen, L.D., Musovic, S., Nesme, J., Sørensen, S.J., 2021. Broad dissemination of plasmids across groundwater-fed rapid sand filter microbiomes. mBio 12 (6), e03068, 21. Pretty, J., 2018. Intensification for redesigned and sustainable agricultural systems. Science. https://doi.org/10.1126/science.aav0294. Pretty, J., Bharucha, Z.P., 2015. Integrated pest management for sustainable intensification of agriculture in Asia and Africa. Insects 6, 152–182. https://doi.org/ 10.3390/insects6010152. Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J., Glöckner, F.O., 2012. The SILVA ribosomal RNA gene database project: improved data processing and web-based tools. Nucleic Acids Res. 41 (D1), D590–D596. Rios Miguel, A.B., Jetten, M.S.M., Welte, C.U., 2020. The role of mobile genetic elements in organic micropollutant degradation during biological wastewater treatment. Water Res. X https://doi.org/10.1016/j.wroa.2020.100065. Rolph, C.A., Jefferson, B., Brookes, A., Hassard, F., Villa, R., 2020. Achieving drinking water compliance levels for metaldehyde with an acclimated sand bioreactor. Water Res. 184 https://doi.org/10.1016/j.watres.2020.116084. Rolph, C.A., Jefferson, B., Hassard, F., Villa, R., 2018. Metaldehyde removal from drinking water by adsorption onto filtration media: mechanisms and optimisation. Environ. Sci. 4, 1543–1552. https://doi.org/10.1039/c8ew00056e. Rolph, C.A., Villa, R., Jefferson, B., Brookes, A., Choya, A., Iceton, G., Hassard, F., 2019. From full-scale biofilters to bioreactors: engineering biological metaldehyde removal. Sci. Total Environ. 685, 410–418. https://doi.org/10.1016/j. scitotenv.2019.05.304. Rosenqvist, T., Chan, S., Ahlinder, J., Salomonsson, E.N., Suarez, C., Persson, K.M., Rådström, P., Paul, C.J., 2024. Inoculation with adapted bacterial communities promotes development of full scale slow sand filters for drinking water production. Water Res. 253, 121203. Samuelsen, E.D., Badawi, N., Nybroe, O., Sørensen, S.R., Aamand, J., 2017. Adhesion to sand and ability to mineralise low pesticide concentrations are required for efficient bioaugmentation of flow-through sand filters. Appl. Microbiol. Biotechnol. 101, 411–421. https://doi.org/10.1007/s00253-016-7909-6. Schönenberger, U.T., Simon, J., Stamm, C., 2022. Are spray drift losses to agricultural roads more important for surface water contamination than direct drift to surface waters? Sci. Total Environ. 809 https://doi.org/10.1016/j.scitotenv.2021.151102. Schostag, M.D., Gobbi, A., Fini, M.N., Ellegaard-Jensen, L., Aamand, J., Hansen, L.H., Muff, J., Albers, C.N., 2022. Combining reverse osmosis and microbial degradation for remediation of drinking water contaminated with recalcitrant pesticide residue. Water Res. 216 https://doi.org/10.1016/j.watres.2022.118352. Sekhar, A., Horemans, B., Aamand, J., Sørensen, S.R., Vanhaecke, L., Bussche, J. vanden, Hofkens, J., Springael, D., 2016. Surface colonization and activity of the 2,6- dichlorobenzamide (BAM) degrading aminobacter sp. strain MSH1 at macro- and micropollutant BAM concentrations. Environ. Sci. Technol. 50, 10123–10133. https://doi.org/10.1021/acs.est.6b01978. Simms, L.C., Dawson, J.J.C., Paton, G.I., Wilson, M.J., 2006. Identification of environmental factors limiting plant uptake of metaldehyde seed treatments under field conditions. J. Agric. Food Chem. 54, 3646–3650. https://doi.org/10.1021/ jf060231a. Sørensen, S.R., Holtze, M.S., Simonsen, A., Aamand, J., 2007. Degradation and mineralization of nanomolar concentrations of the herbicide dichlobenil and its persistent metabolite 2,6-dichlorobenzamide by Aminobacter spp. isolated from dichlobenil-treated soils. Appl. Environ. Microbiol. 73, 399–406. https://doi.org/ 10.1128/AEM.01498-06. Sun, W., Jing, Z., 2023. Migration of rare and abundant species, assembly mechanisms, and ecological networks of microbiomes in drinking water treatment plants: effects of different treatment processes. J. Hazard Mater. 457, 131726. Taylor, A.C., Mills, G.A., Gravell, A., Kerwick, M., Fones, G.R., 2022. Pesticide fate during drinking water treatment determined through passive sampling combined with suspect screening and multivariate statistical analysis. Water Res. 222 https:// doi.org/10.1016/j.watres.2022.118865. Toccalino, P.L., Gilliom, R.J., Lindsey, B.D., Rupert, M.G., 2014. Pesticides in groundwater of the United States: decadal-scale changes, 1993-2011. Ground Water 52, 112–125. https://doi.org/10.1111/gwat.12176. Trikannad, S.A., van Halem, D., Foppen, J.W., van der Hoek, J.P., 2023. The contribution of deeper layers in slow sand filters to pathogens removal. Water Res. 237 https:// doi.org/10.1016/j.watres.2023.119994. UKWIR, 2015. Pesticide Risk Mapping and Catchment Interventions - Phase 1. Vandermaesen, J., Du, S., Daly, A.J., Baetens, J.M., Horemans, B., de Baets, B., Boon, N., Springael, D., 2022. Interspecies interactions of the 2,6-dichlorobenzamide degrading aminobacter sp. MSH1 with resident sand filter bacteria: indications for mutual cooperative interactions that improve BAM mineralization activity. Environ. Sci. Technol. 56, 1352–1364. https://doi.org/10.1021/acs.est.1c06653. Vargas, R., Hattori, T., 1986. Protozoan predation of bacterial cells in soil aggregates. FEMS Microbiol. Ecol. 38, 233–242. Wang, Q., Hou, J., Yuan, J., Wu, Y., Liu, W., Luo, Y., Christie, P., 2020. Evaluation of fatty acid derivatives in the remediation of aged PAH-contaminated soil and microbial community and degradation gene response. Chemosphere 248, 125983. Zhang, R., Cui, Z., Zhang, X., Jiang, J., Gu, J.D., Li, S., 2006. Cloning of the organophosphorus pesticide hydrolase gene clusters of seven degradative bacteria isolated from a methyl parathion contaminated site and evidence of their horizontal gene transfer. Biodegradation 17, 465–472. https://doi.org/10.1007/s10532-005- 9018-6. Zhang, W., Li, J., Zhang, Y., Wu, X., Zhou, Z., Huang, Y., Zhao, Y., Mishra, S., Bhatt, P., Chen, S., 2022. Characterization of a novel glyphosate-degrading bacterial species, Chryseobacterium sp. Y16C, and evaluation of its effects on microbial communities in glyphosate-contaminated soil. J. Hazard Mater. 432 https://doi.org/10.1016/j. jhazmat.2022.128689. L. Pickering et al. https://doi.org/10.1038/ismej.2014.175 https://doi.org/10.1038/ismej.2014.175 https://doi.org/10.1021/acsestwater.2c00313 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref20 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref20 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref20 https://doi.org/10.3389/fmicb.2016.01692 https://doi.org/10.1021/acs.est.6b05027 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref23 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref23 https://doi.org/10.1016/j.scitotenv.2022.158175 https://doi.org/10.1016/j.biortech.2021.126192 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref26 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref26 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref26 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref26 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref27 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref27 https://doi.org/10.1016/j.watres.2008.10.044 https://doi.org/10.1016/j.watres.2008.10.044 https://doi.org/10.1016/j.jhazmat.2020.124457 https://doi.org/10.1016/j.scitotenv.2018.11.260 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref31 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref31 https://doi.org/10.1016/j.jhazmat.2011.08.036 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref33 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref33 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref33 https://doi.org/10.1126/science.aav0294 https://doi.org/10.3390/insects6010152 https://doi.org/10.3390/insects6010152 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref36 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref36 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref36 https://doi.org/10.1016/j.wroa.2020.100065 https://doi.org/10.1016/j.watres.2020.116084 https://doi.org/10.1039/c8ew00056e https://doi.org/10.1016/j.scitotenv.2019.05.304 https://doi.org/10.1016/j.scitotenv.2019.05.304 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref41 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref41 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref41 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref41 https://doi.org/10.1007/s00253-016-7909-6 https://doi.org/10.1016/j.scitotenv.2021.151102 https://doi.org/10.1016/j.watres.2022.118352 https://doi.org/10.1021/acs.est.6b01978 https://doi.org/10.1021/jf060231a https://doi.org/10.1021/jf060231a https://doi.org/10.1128/AEM.01498-06 https://doi.org/10.1128/AEM.01498-06 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref48 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref48 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref48 https://doi.org/10.1016/j.watres.2022.118865 https://doi.org/10.1016/j.watres.2022.118865 https://doi.org/10.1111/gwat.12176 https://doi.org/10.1016/j.watres.2023.119994 https://doi.org/10.1016/j.watres.2023.119994 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref52 https://doi.org/10.1021/acs.est.1c06653 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref54 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref54 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref55 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref55 http://refhub.elsevier.com/S0045-6535(24)01850-2/sref55 https://doi.org/10.1007/s10532-005-9018-6 https://doi.org/10.1007/s10532-005-9018-6 https://doi.org/10.1016/j.jhazmat.2022.128689 https://doi.org/10.1016/j.jhazmat.2022.128689 How bioaugmentation for pesticide removal influences the microbial community in biologically active sand filters 1 Introduction 2 Materials and methods 2.1 Chemicals and water for pilot sand filter experiments 2.2 Analytical methods for metaldehyde 2.3 Bacterial strains used for bioaugmentation 2.4 Trials of bioaugmentation strains in pilot-scale flow through SSF 2.5 DNA extractions from SSF media 2.6 16S rRNA gene amplicon sequencing for microbial community analysis 2.7 Quantification of total bacteria and known metaldehyde-degrading genes 2.8 Statistical analysis 3 Results and discussion 3.1 Water quality and SSF performance during bioaugmentation 3.2 Impact of bioaugmentation on microbial community in SSF 3.3 Activity of bioaugmentation agents 3.4 Why do some bioagumentation agents perform better than others? 3.5 Health implications of biological pesticide degradation 3.6 Application of bioaugmentation in drinking water treatment 4 Conclusion Data access CRediT authorship contribution statement Declaration of competing interest Data availability Acknowledgments Appendix A Supplementary data References