All samples were collected by the RAS biofilter (UWM biofilter) at the University of Wisconsin-Milwaukee Great Lakes Aquaculture Facility. Measured from the base, the biofilter is ~2.74 m high and ~1.83 m in diameter. The water level within the biofilter is ~2.64 m from the base, with the fluidized sand filter matrix extending to a height of Extends ~1.73 m from the base. The biofilter is filled with Wedron 510 silica sand, which is fluidized to ~200% starting sand volume through the use of 19 Plan 40 PVC probes, each 3.175 cm in diameter. The probes receive inflow from the solid waste clarifier, which rises through the filter matrix. Samples for this study were collected at three depths within the fluidized sand biofilter, defined as surface (~1.32–1.42 m from the biofilter base), middle (~0.81–0.91 m from the biofilter base), and bottom (~0.15–0.30 m, made from biofilter base). Images of the UWM biofilter and sampling locations are shown in Figure 1  . The maximum flow rate of the biofilter inflow is 757 L per minute, resulting in a hydraulic retention time of ~9.52 min. Typical system water quality parameters are as follows (mean ± standard deviation): pH 7.01 ± 0.09, oxidation-reduction potential 540 ± 50 (mV), water temperature 21.7 ± 0.9 (°C), and dissolved oxygen (DO) of biofilter effluent 8.20 ± 0.18 mg/l. The biofilter is designed for maximum operation at 10 kg of feed per day, which is based on predicted ammonia production from fish protein breakdown at this feeding rate (  Timmons and Ebeling, 2013  ).

FIGURE 1. ILLUSTRATION OF UW-MILWAUKEE'S RECYCLING AQUACULTURE SYSTEM (RAS) FLUID SAND BIOFILTER  . For illustrative purposes only a single inflow pipe is shown. Nineteen of these pipes are present in the system. Water flow is shown with directional arrows, sample locations are marked with circles, and biofilter elevation is listed.

Sample collection, processing and DNA extraction

Samples from the top of the biofilter matrix were collected in autoclaved 500 mL polypropylene bottles. Two samples from the surface of the biofilter were collected during the last 2 months of a yellow perch rearing cycle and then immediately before the start of a new rearing cycle in the system. After the system was stocked with fish, samples were collected approximately every week for the first half of the new rearing cycle (the yellow perch strains present during this study take approximately 9 months to grow to market size). After collection, water from the biofilter matrix samples was decanted into a second sterile 500 mL bottle for further processing. Then, approximately 1 g of wet weight sand was removed from the sample bottle and frozen at −80 °C for storage prior to DNA extraction. Water samples were filtered to 0. 22 μm filters (47 mm mixed cellulose esters, EMD Millipore, Darmstadt, Germany) frozen at −80 °C and macerated with a sterilized spatula before DNA extraction. To address the spatial distribution of bacterial taxa separately, depth samples were collected from the filter matrix using 50 ml syringes with attached weighted Tygon tubing (3.2 mm ID, 6.4 mm OD; Saint-Gobain SA, La Défense, Courbevoie , France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA and macerated with a sterilized spatula before DNA extraction. To address the spatial distribution of bacterial taxa separately, depth samples were collected from the filter matrix using 50 ml syringes with attached weighted Tygon tubing (3.2 mm ID, 6.4 mm OD; Saint-Gobain SA, La Défense, Courbevoie , France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA and macerated with a sterilized spatula before DNA extraction. To address the spatial distribution of bacterial taxa separately, depth samples were collected from the filter matrix using 50 ml syringes with attached weighted Tygon tubing (3.2 mm ID, 6.4 mm OD; Saint-Gobain SA, La Défense, Courbevoie , France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. Depth samples were extracted from the filter matrix using 50 mL syringes with attached weighted Tygon tubing (3.2 mm ID, 6.4 mm OD; Saint-Gobain SA, La Défense, Courbevoie, France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. Depth samples were extracted from the filter matrix using 50 mL syringes with attached weighted Tygon tubing (3.2 mm ID, 6.4 mm OD; Saint-Gobain SA, La Défense, Courbevoie, France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA 4mm OD; Saint-Gobain SA, La Défense, Courbevoie, France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA 4mm OD; Saint-Gobain SA, La Défense, Courbevoie, France). Samples were categorized according to the approximate distance from the filter base as surface, center, and bottom. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA. The tubing was sterilized with 10% bleach and rinsed three times with sterile deionized water between samplings. DNA was extracted separately from biofilter sand and water samples (~1 g wet weight and 100 mL, respectively) using the MP Bio FastDNA^®  SPIN Kit for Soil (MP Bio, Solon, OH, USA) according to the manufacturer's instructions, except that each sample was beaten for 2 minutes with the beads included in the MP Bio FastDNA ® SPIN Kit at the single operating speed of the Mini-BeadBeater  ^-  16 (Biospec Products, Inc., Bartlesville, OK, USA). DNA quality and concentration were checked using a NanoDrop  ^®  Lite (Thermo Fisher Scientific Inc., Waltham, MA, USA). Sample details and associated environmental data and molecular analyzes are presented in Table S1.

Ammonia and nitrite measurements

For both the time series and depth profiles, a Seal Analytical AA3 Autoanalyzer (Seal Analytical Inc., Mequon, WI, USA) was used to quantify ammonia and nitrite using the manufacturer-supplied phenol and sulfanilamide protocols on two separate channels became. To quantify nitrite only, the cadmium reduction column was not installed in the Auto Analyzer. RAS operators recorded all other chemical parameters from submerged probes that measured temperature, pH and oxidation-reduction potential. Following laboratory standard operating procedures, RAS operators used Hach colorimetric kits to measure ammonia and nitrite concentrations in the rearing tank.

16S rRNA gene sequencing

To maximize read depth for a temporal study of biofilter surface communities, we used the Illumina HiSeq platform and separately targeted the V6 region of the 16S rRNA gene for  archaea  and  bacteria  . In total, we received community data from 15 dates for temporal analysis. To interrogate changes in the spatial distribution of taxa across depth in the biofilter and obtain increased taxonomic resolution, we used 16S rRNA gene V4-V5 region sequencing on an Illumina MiSeq. We obtained samples from three depths  n  = 5 for the surface,  n  = 5 for the middle and  n = 4 for the bottom. Example metadata is listed in Table S1. Extracted DNA samples were sent to the Josephine Bay Paul Center at the Marine Biological Laboratory (V6  Archaea  and V6  Bacteria  ; V4-V5 samples from 12/8/2014 to 2/18/2015) and to the Great Lakes Genomic Center (V4-V5 samples from 11/18 .2014, 12/2/2014, 12/18/2014) for massively parallel 16S rRNA gene sequencing using previously published bacterial (  Eren et al., 2013  ) and archaeal (  Meyer et al., 2013  ) V6 Illumina HiSeq and bacterial V4 V5 Illumina MiSeq chemistry (  Huse et al., 2014b  ;  Nelson et al., 2014 ). Reaction conditions and primers for all Illumina runs are listed in the citations above and can be accessed at: https://vamps.mbl.edu/resources/primers.php#illumina. Sequence run processing and quality control for the V6 data set are in  Fisher et al. (2015)  , while CutAdapt was used to trim the V4-V5 data from low quality nucleotides (phred score <20) and primers (  Martin, 2011  ;  Fisher et al., 2015  ). Trimmed reads were merged using Illumina Utils as described previously (  Newton et al., 2015  ). Minimum Entropy Decomposition (MED) was implemented for each dataset to group sequences (MED nodes = operational taxonomic units, OTUs) for sample community composition and diversity analysis ( Eren et al., 2015  ). MED uses the information uncertainty calculated via Shannon entropy at all nucleotide positions of an alignment to divide sequences into sequence-like groups (  Eren et al., 2015  ). The sequence datasets were parsed with the following minimum substantial abundance settings: Bacteria V6, 377; archaeal V6, 123; bacterial V4-V5, 21. The minimum substantial threshold sets the abundance threshold for inclusion of MED nodes (i.e. OTU) in the final data set. Minimum substantial frequencies were calculated by dividing the total number of 16S rRNA gene sequences per data set by 50,000 as suggested in MED Best Practices (sequence counts are listed in Table S2). The Global Alignment for Sequence Taxonomy (GAST) algorithm was used to assign a taxonomy to sequence reads ( Huse et al., 2008  ) and the Visualization and Analysis of Microbial Population Structures (VAMPS;  Huse et al., 2014a  ) website was used for uses data visualization.

Comammox  -amoA-  PCR

To target comammox  Nitrospira amoA  for PCR and subsequent cloning and sequencing,  amoA  nucleotide sequences were obtained from  van Kessel et al. (2015)  and  Daims et al. (2015)  were aligned with MUSCLE (  Edgar, 2004  ). The alignment was imported into EMBOSS to   generate  an amoA consensus sequence ( Rice et al., 2000  ). Primer sequences were identified from consensus using Primer3Plus (Untergasser et al., 2012)  ,  and  the candidates along with those described by  van Kessel et al. (2015), were evaluated from the consensus sequence in SeqMan Pro (DNAStar) using MUSCLE (  Edgar, 2004  ). The  pmoA  forward primer (  Luesken et al., 2011  ) and the candidate primer COM_amoA_1R (this study; Table  1  ) provided the best combination of read length and specificity and were subsequently used to   amplify amoA genes from our samples.

.

 

Construction of a clone library and phylogenetic analysis

Multiple endpoint PCR approaches were used to   examine the nitrifying community composition of the RAS fluidized sand biofilter for amoA  (  Gammaproteobacteria, Betaproteobacteria, Archaea,  and Comammox  Nitrospira  ),  nxrA  (  Nitrobacter  spp.), and  nxrB  (Non-  Nitrobacter NOB). The primer sets and reaction conditions used are   listed in Table 1 . All endpoint PCR reactions were performed with a volume of 25 μl: 12.5 μl 2x Qiagen PCR master mix (Qiagen, Hilden, Germany), 1.5 μl appropriate primer mix (F&R), 0.5 μl bovine serum albumin ( BSA), 0.75 μl 50 mM MgCl  ₂  and 1 μl DNA extract.

DNA samples of biofilter water and sand from four different time points of the rearing cycle were used to create clone libraries of  archaeal amoA  and  Nitrospira  sp. nxrB  .  A sample from the middle of the sand biofilter was used to construct clone libraries for Betaproteobacteria  -amoA  and Comammox  -amoA . The middle biofilter sample was selected because it produced well-defined amplicons suitable for cloning target  amoA genes. All PCR reactions for cloning libraries were constructed using a TOPO-PCR 2.1 TA cloning kit plasmid (Invitrogen, Life Technologies, Carlsbad, CA). Libraries were sequenced on an ABI 3730 Sanger sequencer with M13 forward primers. Vector plasmid sequence contamination was removed using DNAStar (Lasergene Software, Madison, WI).

Cloned sequences of  Betaproteobacteria amoA, Archaea amoA,  and  Nitrospira nxrB  from this study were added to ARB alignment databases from previous studies (  Abell et al., 2012  ;  Pester et al., 2012  ,  2014  ). Comammox  -amoA  sequences from this study were compared with those from  van Kessel et al. matched. (2015)  ,  Pinto et al. (2015)  and  Daims et al. (2015)  using MUSCLE and imported into a new ARB database where the alignment was heuristically corrected prior to phylogenetic tree reconstruction. For the AOA, AOB and  Nitrospira amoA phylogenies, relationships were determined using maximum likelihood (ML) with RAxML on the Cipres Science Gateway (  Miller et al., 2010  ;  Stamatakis, 2014  ) and Bayesian inference (BI). calculated by MrBayes with a significant posterior probability of <0.01 and the associated consensus tree (  Abell et al., 2012  ;  Pester et al., 2012  ,  2014  ) integrated by ARB into a tree block within the input nexus file to reduce the calculation time (  Miller et al., 2010  ;  Ronquist et al., 2012  ). Consensus trees were then calculated from the ML and BI reconstructions using ARB's consensus tree algorithm (  Ludwig et al., 2004  ).

The  Nitrospira nxrB  sequences generated in this study were significantly shorter than those used for the  nxrB phylogenetic  reconstruction in  Pester et al. (2014)  , therefore we did not perform phylogenetic reconstructions as with the other marker genes. Instead, the UWM Biofilter and  Candidatus  Nitrospira nitrificans sequences were added to the majority consensus tree by  Pester et al. (2014)  using the quick-add parsimony tool of the ARB package (  Ludwig et al., 2004  ). This tool uses sequence similarity to add sequences to pre-existing trees without changing the tree topology.

 

qPCR assays for target marker genes

Quantitative PCR assays were developed to   distinguish two Nitrospira nxrB  genotypes and two  Nitrosomonas amoA genotypes in our system. Potential qPCR primer sequences were identified using Primer3Plus (  Untergasser et al., 2012  ) on MUSCLE (  Edgar, 2004 ) generated alignments in DNAStar (Lasergene Software, Madison, WI). Primer concentrations and annealing temperatures were optimized for specificity for each reaction target. Primers were checked using Primer-BLAST on NCBI to ensure assays matched their target genes. The newly designed primers were tested for cross-reactivity between genotypes using the non-target genotype sequence in both endpoint and real-time PCR dilution series. After optimization, all assays amplified only the target genotype. Due to the high sequence similarity between the two  archaeal amoA  genotypes (>90% identity) in our system, a single qPCR assay was developed to target both genotypes using the steps described above. The two closely related sequence types were pooled in equimolar amounts for reaction standards. A Komammox amoA  qPCR primer set was developed using the same methods as the other assays presented in this study. All test conditions are listed in Table  1  . All qPCR assays were performed on an Applied Biosystems StepOne Plus thermal cycler (Applied Biosystems, Foster City, CA).  Cloned target genes were used to generate standard curves from 1.5 × 10  ^{6 to 15 copies per reaction.} All reactions were performed in triplicate, with melting curve and endpoint confirmation of the assays (qPCR standard curve parameters and efficiency are listed in Table S3).

 

Statistics and data analysis

Taxonomy-based data were visualized with heatmaps created in the R statistical language (  R Core Team, 2014  ) by implementing functions from the gplots, Heatplus from Bioconductor Lite, VEGAN, and RColorBrewer libraries. MED nodes were used in all sample diversity metrics. The EnvFit function in the VEGAN (  Oksanen et al., 2015  ) R package was used to test the relationship between RAS observation data and changes in biofilter bacterial community composition. Pearson's correlations were calculated using the Hmisc package in R (  Harrell, 2016  ) to test whether 16S rRNA,  amoA  and  nxrB  gene copies were correlated over time. Kruskal-Wallis rank sum tests were performed in the R basic statistics package ( R Core Team, 2014  ) to test whether the populations of the above genes were stratified by depth. VEGAN's ADONIS function was used on the V4-V5 depth dataset to test the significance of the observed Bray-Curtis dissimilarity as a function of the categorical factors of depth, with strata = ZERO because the same biofilter was sampled multiple times.

 

Biomass model

To determine whether the observed ammonia removal could provide the energy required to support the number of potential ammonia-oxidizing microorganisms (AOM) in the biofilter as quantified via qPCR, we included the steady-state biomass concentration from the measured ammonia oxidation modeled by the following equation:

: 

 X  _AO  is  defined in previous models (Mußmann et al., 2011) as the biomass concentration of ammonia oxidizers in milligrams per liter, but in this study we converted to cells per wet gram of sand by finding the mean grams of sand per liter of water in Biofilter. Θ  _x  is the mean cell residence time (MCRT) in days and was unknown for the system. Θ is the hydraulic residence time in days, which in this system is ~9.52 min or 0.0066 days. Y  _AO  is the growth yield of ammonia oxidizers and  b  _AO  is the endogenous respiration constant of ammonia oxidizers, which  ^{were estimated to be 0.34 kg volatile suspended solids (VSS)/kg NH4 +}  -N and 0.15 d  -1 by  Mußmann et al. (2011)  . Δ  S  _NH  3  is the change in substrate ammonia concentration between inlet and outlet in mg/l. To  calculate  _{_}  _  _  _  _  _ *C/μm  ³  (  Mußmann et al., 2011  ) to relate the biovolume to endogenous respiration. The modeled biomass concentration was plotted against a range of potential MCRT for a RAS fluidized sand filter (Summerfelt, personal communication). The results of all amoA  qPCR assays were combined to estimate total ammonia-oxidizing microorganism biomass in copy numbers per gram wet weight of sand. The modeled biomass was then compared to our AOM qPCR assay results. A commented R script for the model is available on GitHub ( https://github.com/rbartelme/BFprojectCode.git  ).

 

NCBI sequence accession numbers

Bacterial V6, V4-V5, and archaeal V6-16S rRNA gene sequences generated in this study are available from the NCBI SRA (SRP076497; SRP076495; SRP076492). Partial gene sequences for amoA and nxrB are available through NCBI Genbank and have accession numbers KX024777–KX024822. Results Biofilter chemistry results RAS operational data were examined from the beginning of a yellow perch rearing cycle to approximately 6 months thereafter. The mean biofilter influent concentrations of ammonia and nitrite were 9.02 ± 4.76 and 1.69 ± 1.46 μM, respectively. Biofilter wastewater ammonia concentrations (3.84 ± 7.32 μM) remained within the toxicological limitations (

< 60 μM) of P. flavescens grown in the system. Occasionally, nitrite accumulated above the recommended threshold of 0.2 μM in both the rearing tank (0.43 ± 0.43 μM) and the biofilter effluent (0.73 ± 0.49 μM). No major fish diseases were reported during the RAS's operational period. Environmental and operational data are listed in Table S1. Bacteria and archaea accumulations within the biofilter Characterization of the RAS biofilter bacterial community showed that both the sand-associated and aquatic communities were diverse at a broad taxonomic level; Seventeen phyla averaged >

0.1% in each of the biofilter sand and water bacterial communities (see Table S2 for an example taxonomic characterization of the genus). Proteobacteria (on average 40% of biofilter sand community sequences and 40% of water sequences) and Bacteroidetes (18% in sand, 33% in water) dominated both water and sand bacterial communities. In taxonomic classification at the family level, the community associated with biofilter sand was different from the aquatic community. The majority of sequences in the sand samples were assigned to the bacterial groups Chitinophagaceae (average relative abundance 12%), Acidobacteria family unknown (9%), Rhizobiales family unknown (6%), Nocardioidaceae (4%), Spartobacteria family unknown ( 4%) and Xanthomonadales - family unknown (4%). Water samples were dominated by sequences classified as Chitinophagaceae (14%), Cytophagaceae (8%), Neisseriaceae (8%), and Flavobacteriaceae (7%). At the genus level , Kribbella, Chthoniobacter, Niabella, and Chitinophaga were the most numerous taxa classified, each with an average of >3% relative abundance in the biofilter samples. Using Minimum Entropy Decomposition (MED) to obtain highly discriminatory sequence classification, we identified 1261 nodes (OTUs) in the entire bacterial dataset. A MED-based comparison of bacterial community composition (Figure 1 ) supported the patterns observed using a broader taxonomic classification, indicating that the biofilter sand-associated community was distinct from the assemblage present in the biofilter water. In contrast to the high diversity in the bacterial community, we found that the archaeal community is dominated by a single taxonomic group belonging to the genus Nitrososphaera This taxon accounted for >99.9% of the Archaea -classified sequences identified in the biofilter samples (Table S2). This taxon was also almost entirely represented by a single sequence (>95% of sequences classified as Archaea ) that was identical to a number of Thaumarchaeota sequences deposited in the database, including the complete genome of Candidatus Nitrosocosmicus oleophilus (CP012850), together with clones from activated sewage sludge, wastewater treatment and freshwater aquariums (KR233006, KP027212, KJ810532–KJ810533). Initial characterization of the biofilter community composition revealed distinct communities between the biofilter sand and decanted biofilter water (Figure 2 ). Based on these data and the fact that fluidized bed biofilter nitrification occurs primarily in particle-bound biofilms ( Schreier et al., 2010 ), we focused our further analyzes on the biofilter sand matrix. In the sand samples, we observed a significant change in bacterial community composition (MED nodes) over time (Table 2 ). The early part of the study, which included a period when market size yellow perch were present in the system (sample −69 and −26), a fallow period after fish removal (sample 0), and the period after restocking of mixed juveniles (Samples 7 and 14) had a more variable bacterial community composition (mean Bray-Curtis similarity 65.2 ± 6.5%) than the remaining samples ( n = 9), which were collected at time points after an adult food source had been started (20.0 ± 6.4%, Figure 3 ). Several operational and measured physical and chemical parameters, including oxidation-reduction potential, feed size, conductivity, and nitrite from the biofilter, were correlated ( S < 0.05) with the time-dependent changes in bacterial community composition (see Table 2 for environmental correlation results).

 

.

	FIGURE 2. DENDROGRAM ILLUSTRATING THE BACTERIAL COMMUNITY COMPOSITIONAL RELATIONSHIPS BETWEEN BIOFILTER SAND AND BIOFILTER WATER SAMPLES. A complete linkage dendrogram is represented using Bray-Curtis sample dissimilarity relationships based on nodal distributions of minimum entropy decomposition between samples (V6 dataset). The leaves of the dendrogram are labeled with the day count, with 0 representing the start of a fish rearing cycle. Negative numbers are days before a new breeding cycle. The day count is followed by the date of sampling (MM.DD.YY). See Table S1 for example metadata.
	TABLE 2. CORRELATIONS BETWEEN ENVIRONMENTAL VARIABLES AND BACTERIAL COMMUNITY COMPOSITION.
	FIGURE 3. NON-METRIC MULTIDIMENSIONAL SCALING DIAGRAM OF BRAY-CURTIS BACTERIAL COMMUNITY COMPOSITION DISsimilarity BETWEEN SAMPLING POINTS. nMDS stress = 0.07 and dimensions (k) = 2. Arrows show the progression of the sample through time from the end of a rearing cycle (day number -69 and -26) to a period without fish (0) and into the subsequent rearing cycle ( 7-126). The circle shows samples taken after the fish had grown to a size where feed type and amount were stabilized (3 mm pelleted feed and 3-7 kg feed per day).
Using a second sequence data set (V4-V5 16S rRNA gene sequences), we examined sand-associated bacterial community composition across a depth gradient (surface, middle, bottom). We found that the bacterial communities in the top sand samples were different from those in the middle and bottom (ADONIS  R  2  = 0.74,  p  = 0.001; Figure  4  ). The  planctomycetes  represented a larger proportion of the community in the surface sand (on average 15.6% of the surface sand versus 9.6% of the middle/lower sand), while the middle and lower layers harbored a larger proportion of  Chitinophagaceae  (7.4% in the surface sand vs. 16.8% middle/bottom) and  Sphingomonadaceae (2.4% in the surface vs. 7.9% in the middle/bottom; Figure  4  ).
	FIGURE 4. DEPTH COMPARISON OF BACTERIAL BIOFILTER COMMUNITY COMPOSITION. A heatmap is presented for all bacterial families with a relative abundance of ≥ 1% in each sample. Relative taxon abundance was generated from V4–V5 16S rRNA gene sequencing and is presented on a scale of 0 to 25%. The dendrogram represents the Bray-Curtis dissimilarity between sample community composition. Sample IDs are listed and sample depth is indicated by on the graph next to the dendrogram. Sample names correspond to sample metadata in Table S1.

 

Nitrification community composition and phylogeny

The massively parallel 16S rRNA gene sequencing data showed that bacterial taxa not associated with nitrification comprised the majority (~92%) of the sand biofilter bacterial community. In contrast, >99.9% of archaeal 16S rRNA gene sequences were assigned to a single taxon associated with known AOA. Among bacterial taxa, Nitrosomonas represented 1% of the total community in all samples, and no Nitrobacter sequences were obtained. We were also unable to amplify Nitrobacter nxrA genes (Figure S1) with a commonly used primer set ( Poly et al., 2008 ; Wertz et al., 2008 ). In contrast, Nitrospira was quite abundant, accounting for 2–5% of the total bacterial community (Table S2).

 

---

Context: 

 

ID: 425

 

Biofilms consist of a mucous layer (a film) in which mixed populations[1] of microorganisms (e.g. bacteria, algae, fungi, protozoa) in concentrations of 1012 cells per milliliter of biofilm[1] and of multicellular organisms[1] such as Rotifers, nematodes, mites, bristles or insect larvae that feed on the microorganisms are embedded. In everyday life, they are often perceived as a slippery, soft-feeling, water-containing layer of mucus or coating. Other colloquial names are growth, Kahmhaut or Sielhaut.

By Asw-hamburg - Own work, CC BY-SA 4.0, https://commons.wikimedia.org/w/index.php?curid=46898752

Description

Biofilms predominantly form in aqueous systems when microorganisms settle on interfaces. In principle, all surfaces can be covered by biofilms: between gas and liquid phases (e.g. free water level), liquid and solid phases (e.g. gravel on the bottom of the water) or between different liquid phases (e.g. oil droplets in the water ). The interface on which the biofilm forms, or more precisely the phase into which the film does not or hardly grows, forms the substratum (substrate; that which extends underneath).

In the broader sense, biofilm refers to all aggregates of microorganisms that are embedded in a layer of mucus that they form.[2] Suspended solids in water often consist of mineral particles covered by biofilms. The activated sludge in sewage treatment plants also has essential biofilm properties. It consists of flakes that themselves have a surface suitable for colonization. Biofilms can be considered a very primitive form of life because the oldest fossils that have been found so far come from microorganisms in biofilms that lived 3.2 billion years ago. These are stromatolites (biogenic sedimentary rocks) found in Western Australia (Pilbara Craton). Biofilm as a form of life has proven itself so well that it is still widespread today. The vast majority of microorganisms live in nature in the form of biofilms.[4][FSE 1]

Composition

Fig. 2: Macromolecules of a biofilm. (Modified according to Fuchs[FSE 2]) From above:
Cytoplasm (CP) of a spheroplasted bacterium with cytoplasmic membrane (CPM).
Intercellular (IC) glycocalyx with exo-polysaccharides (EPS), DNA (DNA), hydrophobic (HPr) and water-soluble proteins (SPr).
Periplasmic membrane (PPM), cell wall (W), periplasm (PPl), cytoplasmic membrane and cytoplasm of a bacterium.

Apart from the microorganisms, the biofilm mainly contains water. Extracellular polymeric substances (EPS) secreted by the microorganisms combine with water to form hydrogels, creating a mucous-like matrix in which nutrients and other substances are dissolved. Inorganic particles or gas bubbles are often trapped in the matrix. Depending on the type of microorganisms, the gas phase can be enriched with nitrogen, carbon dioxide, methane or hydrogen sulfide.

The EPS consist of biopolymers that are able to form hydrogels and thus give the biofilm a stable shape. This involves a wide spectrum of polysaccharides, proteins, lipids and nucleic acids (extracellular DNA).

Different types of microorganisms usually live together in biofilms. In addition to the original biofilm formers, other single-celled organisms (amoebas, flagellates, etc.) can also be integrated. Aerobic and anaerobic zones can occur a few hundred micrometers apart, allowing aerobic and anaerobic microorganisms to live close together.

Shape

Fluorescence microscopic image of a multi-species biofilm on stainless steel.
In the core area, the biofilm is usually compact (basic biofilm). The edge area (surface biofilm) can either be compact and regularly shaped and form a flat interface to the fluid flowing over it, or it can be blurred and much looser. In the latter case, the surface biofilm can resemble a mountain-and-valley path if, for example, bacterial species grow into the fluid in a thread-like manner (filamentous) or if the substrate is populated with protozoa (e.g. bellworms) or higher types of organisms.

The biofilm matrix is ​​then often permeated by pores, caverns and passages, which enable material exchange between the bacterial cells and a supply of water. Mushroom-shaped or tower-like structures are often found. Convective mass transport processes occur there when liquid flows through them. In the area of ​​the surface of the biofilm, convective mixing processes can also be triggered by the movement of outgrowths protruding into the flow (e.g. “wastewater fungi” such as Sphaerotilus natans). Inside biofilms, dissolved substances are transported primarily through diffusion. Cells or entire parts of the biofilm can repeatedly be released at the boundary layer with the water and be absorbed by the water flowing past.

Formation and maturation of biofilms

Fig. 4: Phases and microscopic images of biofilm development. The emergence and formation of a biofilm can be divided into three phases: the induction phase (Figs. 4 and 6, 1–2), the accumulation phase (3) and the existence phase (4–5).

Colonization of surfaces

According to popular belief, typical microorganisms have flagella (Fig. 6, 1) and move freely in the water column. In fact, such swarmer cells [FSE 3] are usually only the dispersal stage of biofilm inhabitants.

There is a compelling reason why the absolute majority of bacteria and archaea are rooted in biofilms: otherwise they would be washed out of their biotope by the water necessary for life. Soil bacteria would end up in the nearest river and from there begin their final journey into the sediment of an ocean. The same would happen to the microorganisms in the activated sludge from sewage treatment plants.

In order to be able to leave the free water at all, microorganisms need water-repellent hydrophobic substances on the surface of their cells. These enable organisms to attach to hydrophobic surfaces based on van der Waals forces. Since almost all areas in aquatic biotopes are covered with biofilms[FSE 4], most swarmer cells associate with existing biofilms.

However, such organisms can also attach themselves directly to unpopulated areas. Smooth hydrophobic surfaces, such as B. polystyrene or the cuticle of many plants can be colonized directly, but only if they can be wetted with water. However, thanks to the lotus effect, many plants avoid the growth of microorganisms on their leaves.

A thin, viscous layer of organic substances initially accumulates on empty hydrophilic surfaces. These biopolymers originate from the mucous membranes that form around bacterial cells (EPS), occasionally detach completely or partially and become adsorptively bound upon contact with interfaces. Such biogenic substances are omnipresent in nature.[FSE 5]

The metamorphosis into a biofilm inhabitant

Fig. 5: Life cycle of Caulobacter. A swarmer cell (1) sheds its flagella and the pili are shortened (2). The resulting stalk cell (3) grows and forms new swarm cells (4)[FSE 6]

If the site of attachment allows the organism in question to grow, it will usually shed its flagellum(s). However, in many organisms a much deeper change occurs.

This is clearly visible in Caulobacter, an aerobic α-Proteobacterium. After losing the flagellum, the swarmer cell retracts its attachment pili and becomes a stalk cell. In contrast to the swarm cell, it is capable of division and immediately begins with an asymmetrical division. This creates a new swarmer cell. After separation, the stalk cell can repeatedly form new swarmer cells under suitable conditions.[FSE 7]

The changes in the soil bacterium Bacillus subtilis are at least as profound (Fig. 6). After attachment and loss of flagellation, filamentous structures arise during subsequent cell divisions because the cell walls of the organisms are not separated. At the same time, polymers are secreted, which give the resulting film lateral strength. Such changes are triggered epigenetically.[6]

As a result of the proliferation of cells that have attached themselves to a surface, the organisms spread. The interface is initially colonized over the surface in the form of a film (biofilm). At the same time or later, the biofilms grow in multiple layers and ultimately form heterogeneous three-dimensional structures. Up to this phase, Bacillus subtilis produces almost exclusively filamentous cell groups.

Avoidance of competition
There is, in principle, competition for nutrients between the cells of a biofilm, with the cells closest to the food source having a clear advantage. In contrast, the cells inside are in danger of starving. If that happens, they will no longer be able to maintain cohesion. In fact, there are mechanisms of cell density regulation and communication between cells (quorum sensing)[FSE 8] that counteract this.

Such a mechanism was elucidated in detail for the first time in 2015 for Bacillus subtilis.[7] For this purpose, a biofilm from a pure culture of these bacteria was examined in a chemostat bioreactor. The biofilm was continuously supplied with nutrients, and yet the cells periodically stopped growing until the cells inside the biofilm stopped starving. This “oscillation” is based on the following process:

Starving cells inside the biofilm send out a pulse of K+ ions. The biofilm cells of B. subtilis have receptors for these ions, which trigger a whole chain of events. 

All cells, including the well-supplied cells, send out a K+ signal immediately after receiving it. Specific K+ channels exist in the biofilm for the propagation of signals. (Normal diffusion through the polymeric biofilm matrix would be too slow.)
The cells, which are still well supplied, immediately stop their growth, but not their metabolic activity. If there is a nitrogen deficiency, they take e.g. B. glutamine from the nutrient medium, but do not use this amino acid for growth, but split off ammonium from it, which they make available to the biofilm.

If the signals diminish, growth will continue together.[8]
K+-based communication between bacterial cells is not the only one. There are a number of pheromones that can be produced and sensed by organisms. This also initiates the next phase in the existence of a biofilm (see Fig. 6.5). Metamorphosis of cells occurs again. In well-supplied cells, flagellate swarm cells are formed again, whose preferred swimming direction is towards the nutrient source. Many bacteria, like B. subtilis, also form spores in this phase. These are carried by the current and are prepared for long-term nutrient shortages.[FSE 9]

This phase of emigration is by no means the end of a biofilm. For the release of the spores and swarmer cells, the extracellular matrix is ​​only actively dissolved in their surroundings. In the old part of the biofilm, life continues with a new phase of growth.

The fact that the depth extent of the biofilm is limited becomes apparent when entire parts of the biofilm are carried away by the current. Due to the formation of gas bubbles (e.g. due to denitrification and carbon dioxide), the cohesion of biofilm parts is lost. The increase in flow resistance with increasing thickness leads to increased erosion if the biofilm has formed on surfaces subject to flow. Life in such biofilm fragments is not fundamentally different from biofilms that are attached somewhere. Such flakes have all the properties needed to attach to a new surface.

Biocorrosion

Biocorrosion is observed in the presence of biofilms. Here, iron oxidizers contained in the oxygen-loving (aerobic) top layer lead to an attack on the passive layer (of metals) - sulfate reducers existing in the anaerobic layer attach to these points and “eat” into the material.

Microbiologically caused corrosion causes considerable economic damage every year. The proportion of total corrosion (ie abiotic and biotic corrosion) is estimated to be at least 20%; According to more recent findings, it is probably significantly higher. Even higher-alloy materials such as V2A and V4A are damaged. Almost all technical systems are affected: including cooling circuits, water treatment and industrial water systems, energy production in power plants, the production of cars, computers, paint, and the oil and gas industry.[27] In contaminated mining sites, biological leaching of minerals through biofilms leads to large-scale environmental damage to soil, water and air through dust pollution and emissions of sulfuric acid, heavy metals, radon and radionuclides.

Biofouling

In water treatment using membrane processes, biofilms are responsible for biofouling, which leads to serious problems with this technology.

Biofouling also includes biofilms that form on underwater bodies. This can cause significant problems. A biofilm of just a tenth of a millimeter reduces the speed of a tanker by 10 to 15 percent due to increased frictional resistance. This results in increased fuel consumption. In the fight against organic growth (including barnacles and mussels), special substances are painted onto ships, platforms and buoys, the active ingredients of which are released into the water and often pose a significant environmental impact. One such substance is highly toxic tributyltin (TBT), which is now banned worldwide. Also affected are sensor systems for research or monitoring purposes in the maritime sector, where fouling can very quickly lead to functional impairments.

Concentration gradients of physical-chemical parameters in biofilms can be determined using high-resolution microsensors (= functional investigation) and correlated with molecular biological data from the depth distribution of the microbial populations present in the biofilm (= structural investigation). The ideal goal is to combine the structure and function of the microbial populations in the biofilm with (damage/corrosion) data from the growth area. This contributes to a better understanding of the interaction between the damage-causing biofilm and the growth area, which is of particular interest in applied systems (e.g. marine biofilms in steel pipes).

Context: 

Sources include: https://de.wikipedia.org/wiki/Biofilm

1.  Karl Höll:  Water.  ISBN 978-3-110-22677-5 , pp. 663–669 ( limited preview  in Google book search).
2. ↑  Michel Vert, Yoshiharu Doi, Karl-Heinz Hellwich, Michael Hess, Philip Hodge, Przemyslaw Kubisa, Marguerite Rinaudo, François Schué:  Terminology for biorelated polymers and applications (IUPAC Recommendations 2012) . In:  Pure and Applied Chemistry . 84th year, No. 2, 2012, pp. 377–410,  doi : 10.1351/PAC-REC-10-12-04  ( Online  ( Memento  from March 19, 2015) [PDF; accessed on February 10, 2016]) .  Info:  The archive link was inserted automatically and has not yet been checked. Please check the original and archive link according to  the instructions  and then remove this notice.
3. ↑  Andreas Schmidt-Wilckerling:  Metabolic activity of freely suspended and  immobilized  cells of ammonia-oxidizing bacteria.  Diploma thesis, Hamburg (1989).
4. ↑  ^{Jump up to: a  b  c}  Luanne Hall-Stoodley, J. William Costerton and others:  Bacterial biofilms: from the natural environment to infectious diseases . In:  Nature Reviews Microbiology . Vol. 2, No. 2, 2004,  ISSN  1740-1526 ,  PMID 15040259 ,  doi:10.1038/nrmicro821 , pp. 95–108  (PDF file; 0.6 MB) .
5. ↑  Hera Vlamakis, Yunrong Chai, Pascale Beauregard, Richard Losick, Roberto Kolter:  Sticking together: building a biofilm the Bacillus subtilis way . In:  Nat Rev Micro . 11th year, No. 3, 2013, pp. 157–168,  doi : 10.1038/nrmicro2960 .
6. ↑  Yunrong Chai, Thomas Norman, Roberto Kolter, Richard Losick:  An epigenetic switch governing daughter cell separation in Bacillus subtilis . In:  Genes & Development . Volume 24, No. 8, 2010, pp. 754–765,  doi : 10.1101/gad.1915010  ( cshlp.org ).
7. ↑  Jintao Liu, Arthur Prindle, Jacqueline Humphries, Marcal Gabalda-Sagarra, Munehiro Asally, Dong-yeon D. Lee, San Ly, Jordi Garcia-Ojalvo, Gurol M. Suel:  Metabolic co-dependence gives rise to collective oscillations within biofilms . In:  Nature . 523rd volume, No. 7562, 2015, pp. 550–554,  doi : 10.1038/nature14660 .
8. ↑  Arthur Prindle, Jintao Liu, Munehiro Asally, San Ly, Jordi Garcia-Ojalvo, Gurol M. Suel:  Ion channels enable electrical communication in bacterial communities . In:  Nature . Volume 527, No. 7576, 2015, pp. 59–63,  doi : 10.1038/nature15709 .
9. ↑  James A Shapiro:  Thinking about bacterial populations as multicellular organisms . In:  Annual Reviews in Microbiology . Volume 51, No. 1, 1998, pp. 81–104,  doi : 10.1146/annurev.micro.52.1.81  ( annualreviews.org  [PDF]).

EGSB. An extended granular sludge bed reactor (EGSB) is a variant of the UASB concept (Kato et al. 1994). The differentiating feature is that a higher upward flow velocity is provided for the wastewater passing through the sludge bed.

The increased flow allows partial expansion (fluidization) of the granular sludge bed, improving contact between wastewater and sludge and promoting the segregation of small inactive suspended particles from the sludge bed. The increased flow rate is achieved either through the use of high reactors or through wastewater recirculation (or both).

The EGSB concept is suitable for poorly soluble wastewater (less than 1 to 2 g of soluble COD /l) or for wastewater that contains inert or poorly biodegradable suspended solids that must not be deposited in the sludge bed.

Overview of the performance of the reactors. A recent survey (Frankin, 2001) carefully documented 1215 large-scale, high-speed anaerobic reactors built around the world since the 1970s to treat industrial wastewater. The overwhelming majority (72% of all plants) of existing large-scale plants are based on the UASB or EGSB concept developed by Lettinga in the Netherlands. This statistic highlights that the anaerobic granular sludge bed concept is the most successful for scale-up and implementation. The average design load of the UASB of 682 large-scale systems examined was 10 kg COD/m3.d.

Note: COD stands for chemical oxygen demand and refers to the organic matter in wastewater, expressed as the weight of oxygen required to burn it completely. The average design load of the EGSB of 198 large-scale systems examined was 20 kg COD/m3.d. COD removal efficiency largely depends on the type of wastewater; however, the biodegradable COD removal efficiency is generally over 85 or even 90%.

Biodegradable COD is sometimes expressed by the biological oxygen demand ( BOD ) parameter.

The four main applications for high-load anaerobic reactor systems are:

• Breweries and beverage industry
• Distilleries and fermentation industry
• food industry
• pulp and paper.

These four industries together account for 87% of applications. However, the applications of the technology are rapidly expanding, including the treatment of wastewater from the chemical and petrochemical industries, the textile industry, landfill leachate, as well as applications aimed at sulfur cycle conversion and metal removal (see Other Applications). In addition, the UASB concept is also suitable for the treatment of household wastewater in warm climates.

Anaerobic granular sludge bed technology refers to a special reactor concept for the anaerobic treatment of wastewater with high throughput. The concept was introduced with the UASB reactor (UASB = upward-flow anaerobic sludge blanket). A schematic of a UASB reactor is shown in the figure.

From a hardware perspective, at first glance, a UASB reactor is nothing more than an empty tank (i.e. an extremely simple and inexpensive design).

The wastewater is fed into the tank via appropriately arranged inlets. The wastewater flows upward through an anaerobic sludge bed where the microorganisms in the sludge come into contact with the wastewater substrates. The sludge bed consists of microorganisms that naturally form granules (pellets) with a diameter of 0.5 to 2 mm, which have a high sedimentation rate and are therefore not washed out of the system even under high hydraulic loads. The resulting anaerobic degradation process is usually responsible for the production of gas (e.g. biogas containing CH4 and CO2). The upward movement of the released gas bubbles causes hydraulic turbulence, which ensures mixing of the reactor without mechanical parts. At the top of the reactor, the water phase is separated from the sludge solids and gas in a three-phase separator (also called a gas-liquid solids separator). The three-phase separator is usually a gas cap with a settler above it. Baffles are used below the gas cap opening to direct the gas to the gas cap opening.

Brief history of UASB

The UASB procedure was developed by Dr. Gatze Lettinga and colleagues developed it in the late 1970s at Wageningen University (Netherlands). Inspired by publications by Dr. Perry McCarty (Stanford, USA), Lettinga's team experimented with an anaerobic filter concept. The Anaerobic Filter (AF) is a high-speed anaerobic reactor in which biomass is immobilized on an inert porous support material. During experiments with the AF, Lettinga observed that, in addition to the biomass fixed to the carrier material, a large part of the biomass developed into free granular aggregates. The UASB concept crystallized during Gatze Lettinga's trip to South Africa, where he observed the sludge developing into compact granules in an anaerobic wine vinasse treatment plant. The reactor design of the plant visited was a "Clarigestor", which can be considered a precursor to the UASB. The upper part of the "Clarigestor" reactor has a clarifier but no gas cap.

The birth of the UASB

The UASB concept emerged from the realization that an inert support material for biomass attachment is not necessary to maintain a high proportion of active sludge in the reactor. Instead, the UASB concept is based on a high degree of biomass retention through the formation of sludge granules. When developing the UASB concept, Lettinga took into account the need to promote the accumulation of granular sludge and prevent the accumulation of disperse sludge in the reactor. The most important features for the development of granular sludge are, firstly, maintaining an upward flow in the reactor that selects microorganisms to aggregate, and secondly, ensuring adequate separation of solids, liquid and gas to prevent leaching of the sludge grains.

First UASB. The UASB reactor concept was quickly developed into technology, with the first pilot plant installed at a beet sugar refinery in the Netherlands (CSM suiker). Afterwards, a large number of large-scale systems were installed in sugar refineries, potato starch processing plants and other food industries as well as in waste paper factories in the Netherlands. The first publications on the UASB concept appeared in Dutch-language journals in the late 1970s, and the first international publication appeared in 1980 (Lettinga et al. 1980).

Grahik: By Tilley, E., Ulrich, L., Lüthi, C., Reymond, Ph., Zurbrügg, C. - Compendium of Sanitation Systems and Technologies - (2nd Revised Edition). Swiss Federal Institute of Aquatic Science and Technology (Eawag), Duebendorf, Switzerland. ISBN 978-3-906484-57-0., CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=42267210