INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Modern biological treatment of wastewater involves not only C removal, but also elimination of the nutrients P and N (5, 20). This requires the combined or sequential actions of various groups of microorganisms, such as heterotrophic bacteria, phosphate-accumulating organisms (PAO), and nitrifying and denitrifying bacteria. Consequently, purification plants and processes have become increasingly complex to satisfy the needs of the different microorganisms, usually in several reactor stages (5, 27). The integration of different functions in a single reactor would save reaction space and time and therefore is desirable from an economical point of view. However, difficulties often arise in establishing stable nitrification in such complex systems. Nitrifying bacteria (i.e., ammonia-oxidizing bacteria [AOB] and nitrite-oxidizing bacteria [NOB]) usually show low maximum growth rates, relatively low substrate affinities, and high sensitivity to toxic shocks or sudden pH changes (17, 25, 41). In the presence of organic matter, they can be easily outcompeted by heterotrophs for oxygen (56) and ammonia (19). Other problems to be solved are the inhibition of denitrification by the presence of oxygen (5) and the need for cyclic changes of oxic and anoxic (i.e., free of oxygen and nitrate) conditions for biological phosphate removal (34). Biofilm systems are an obvious option for such multifunctional reactors. Slow-growing organisms remain in the reactor by their attached growth; the biofilm matrix might protect bacteria from grazing, harmful substances, or sudden pH shifts; and biofilms can be stratified and therefore provide oxic and anoxic reaction zones (11). During the last 5 years, several studies have addressed nitrifying biofilms through a combination of microsensor measurements and 16S rRNA-based methods, such as fluorescence in situ hybridization (FISH) (39, 48, 50). This approach revealed, e.g., the identity and spatial arrangement of AOB and NOB in various nitrifying systems (39, 48-50) and provided a first estimate of their in situ reaction rates and substrate affinities (48). However, very little is known about how and which nitrifying bacteria are adapted to competition with heterotrophs in more complex systems and how they interact with other processes.

Recently, a biofilm system was proposed that integrates enhanced biological phosphate removal (EBPR) and nitrification and denitrification in a single reactor (3, 18). The biofilm is subjected to a sequencing batch mode, in which an anoxic treatment period is followed by an oxic period to allow for net accumulation of polyphosphate in the biomass, which is removed from the system by backwashing at regular intervals. Substrate balances revealed that the removal of organic carbon and EBPR were successfully combined with nitrogen removal via nitrification and denitrification (3).

In the present study, the microbial ecology of this combined nitrification-EBPR biofilm process was investigated by using microsensor analysis and various molecular techniques, i.e., FISH and analysis of 16S ribosomal DNA (rDNA) and amoA gene sequences. The objectives were to reveal which populations contribute to which part of the process, which AOB and NOB persist under these highly competitive and transient conditions, and how nitrifying activity overlaps, in time and space, with heterotrophic activity, especially EBPR.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Process description. A 20-liter sequencing batch biofilm reactor (SBBR) was established as described previously (18). The artificial wastewater was composed of Na(CH₃COO) x 3H₂O (103 mg liter¹), peptone (200 mg liter¹), (NH₄)₂SO₄ (63 mg liter¹), KH₂PO₄ (44 mg liter¹), KCl (14 mg liter¹), and yeast extract (3 mg liter¹), leading to influent concentrations of 12 mg liter¹ for P and 38 mg liter¹ for N and a chemical oxygen demand (COD) of about 270 mg liter¹. Oxygen and phosphate concentrations in the bulk water were regularly monitored by online measurements with an oxygen electrode (Oxy 196; WTW, Weilheim, Germany) and with a P analyzer (Phosphax Inter; Dr. Lange, Düsseldorf, Germany). Ammonium, nitrate, and COD were determined photometrically with standard test kits (LCK 303, 339, and 314; digital photometer ISIS 6000; Dr. Lange, Düsseldorf, Germany). The lengths of the operation periods were as follows: 20 min of filling (min 0 to 20), 160 min of nonaerated recirculation (min 20 to 180), 260 min of recirculation with aeration (min 180 to 440), and 40 min of draining (min 440 to 480). The process temperature was kept at 20°C. Biofilm was grown on substratum, Kaldnes elements (Purac, Merseburg, Germany): i.e., plastic rings (diameter, 8 mm, height, 8 mm) designed so biofilm could adhere to both the outer surface and the central spaces within the ring. To remove biofilm material with incorporated polyphosphate, the system was backwashed once a week with pressurized air and water. Removal of biomass from the central spaces of the Kaldnes elements was not efficient, leading to complete clogging of these spaces. The preceding FISH analysis of the biofilm structure as described below revealed no visual difference in the composition and spatial organization of the main microbial populations in the upper parts of biofilms originating from the external substratum surface and from the biofilm of clogged central spaces. Therefore, elements filled completely with biogenic material were chosen for microsensor measurements and FISH.

Microsensor measurements. To allow measurements during reactor operation, Kaldnes elements with biofilm were transferred during the initial filling period from the reactor to a separate 750-cm³ flow chamber coupled to the recirculation of the reactor. Microsensors were inserted through small holes in the top lid that were closed with stoppers during the nonaeration period. Vertical concentration microprofiles in the biofilm were measured for oxygen with Clark-type microsensors (45) and for ammonium, nitrite, and nitrate with potentiometric ion-selective microelectrodes of the LIX type (15) as described previously. At least 20 profiles were measured for each parameter at different times covering the whole course of a treatment cycle. Measurements were distributed over four cycles to check for the similarity of conditions on different days of operation.

Rate calculations. Due to the periodic changes in the process conditions, the microprofiles measured in situ do not represent a steady-state situation. Therefore, in the case of nitrate, profiles were corrected to allow the application of a one-dimensional diffusion-reaction model for calculation of specific volumetric rates of net consumption or production. In each nitrate profile, the first value measured in the bulk water was set to t = 0. Then for each depth, the concentration change over time was calculated from two subsequent profiles. The resulting slope was used to correct the time-dependent change of concentration points at t > 0 in each profile by linear interpolation. In that way, every profile was corrected for the dynamics of concentration changes leading to the elimination of transience. The correction procedure just described influenced the absolute values of concentration in the deeper biofilm, but did not change the shapes of the profiles. The correction-of-profile data were not applied to oxygen profiles, because their appearance and the stability of bulk values throughout the oxic period of the process indicated a clear steady-state situation with respect to oxygen.

DNA extraction. DNA was extracted from native biofilm samples stored at 70°C with the FastDNA-Extraction kit for soil (Bio 101, Carlsbad, Calif.), as described in the manufacturer's instructions. The quality of DNA was checked by agarose (1% [wt/vol]) gel electrophoresis.

16S rDNA analysis. A 1-kb fragment of the 16S rDNA gene was amplified with the complement of probe NOLI191 (43) as the forward primer and the unlabeled probe Nso1225 (35) as the reverse primer. The following reaction mixture was used: 50 pmol of each primer, 2.5 µmol of each deoxynucleoside triphosphate (dNTP), 1× PCR buffer, 1 U of SuperTaq DNA polymerase (HT Biotechnology, Cambridge, United Kingdom), and 50 to 100 ng of template DNA. The mixture was adjusted to 100 µl with sterile water. PCR was performed with an Eppendorf Mastercycler (Eppendorf, Hamburg, Germany) with 35 cycles with hot start. The annealing temperature of 56°C for the primer set was determined in previous PCRs run at different temperatures. After checking an aliquot of the PCR product by agarose gel electrophoresis, the DNA was directly ligated into the pGEM-T vector (Promega, Mannheim, Germany) according to the manufacturer's instructions and subsequently transformed to competent high-efficiency Escherichia coli cells (strain JM109; Promega, Mannheim, Germany). White and blue screening was used to screen for recombinant transformants. Inserts of positive clones were analyzed for redundancy by amplified rDNA restriction analysis (ARDRA) (44). One representative clone was sequenced for each distinct fragmentation pattern by Taq Cycle sequencing with the PCR primers or primers M13uni and M13rev on a model ABI 377 sequencer (PE Corporation, Norwalk, Conn.). The sequences were checked for chimera formation with the CHECK_CHIMERA software of the Ribosomal Database Project (32). Sequences were aligned and analyzed by use of the ARB software package (Technical University Munich, Munich, Germany) according to the last release of the 16S rDNA sequence database of December 1998 as well as individually added sequences of recent publications. Trees were calculated for the clones and selected related sequences following the suggestions of Ludwig et al. (31). For tree calculation, maximum-parsimony, distance matrix, and maximum-likelihood methods were used, and the results were combined in a consensus tree.

Probe design. Based on the newly retrieved and published sequences of the Nitrosomonas oligotropha lineage (40), a probe was designed by using the PROBE_DESIGN tool of ARB. The dissociation temperature of the probe was determined by hybridization with a pure culture of Nitrosomonas ureae isolate Nm10 at formamide concentrations ranging from 0 to 60%. Analysis was done by confocal laser-scanning microscopy (CLSM) and image analysis as described elsewhere (13). The specificity of the probe was checked against the ARB database and experimentally tested under the optimal hybridization conditions with the strains listed in Table 1.

FISH. Complete substratum elements with adhering biofilm were fixed with fresh 4% paraformaldehyde solution and alternatively with ethanol, washed with phosphate-buffered saline (PBS), and stored in PBS-ethanol (1:1) at 20°C until further processing (1, 33). After freezing and removal of the plastic material, radial biofilm sections with a thickness of 14 µm were prepared at 18°C, immobilized on gelatin-coated microscope slides, and dehydrated in an ethanol series (50). In situ hybridizations of cells in the biofilm were performed with fluorescently labeled, rRNA-targeted oligonucleotide probes according to the method of Manz et al. (33). The probes and conditions used are listed in Table 2. Probes labeled with the sulfoindocyanine dyes Cy3 and Cy5 were obtained from Interactiva (Ulm, Germany) and Biometra (Göttingen, Germany). In cases in which stringency conditions did not allow simultaneous hybridization with several probes, multiple probe hybridization was performed in subsequent steps by first hybridizing with the probe of higher stringency (58). The biofilm was additionally stained with 4', 6'-diamidino-2-phenylindole (DAPI) after the hybridization step with a solution of 1 mg liter¹ for 10 min or 100 mg liter¹ for 10 s for the purpose of polyphosphate staining (23). Samples were analyzed by standard epifluorescence microscopy on a Zeiss Axioplan II microscope and by CLSM on a Zeiss LSM 510 microscope (Carl Zeiss, Jena, Germany).

Quantification of AOB and NOB. Total AOB abundance was quantified by microscopical counting of cells hybridized with probes Nso1225 and Nso190 in several fields of view along a 50-µm-thick horizontal layer up to a biofilm depth of 600 µm. Distinct populations of AOB were counted after hybridization with probes NEU and NOLI191. Quantification of NOB was done by CLSM as previously described (48); i.e., optical sections with a defined thickness of 0.6 µm were scanned, and by determining the signal area and subsequent multiplication with a volume-specific cell abundance, the cell number for a given volume was calculated. Thorough calibration for NOB quantification was performed by counting DAPI-stained cells in various scanned fields of view (n = 30), leading to volumetric indices with a 95% confidence interval of ±7%. Results with all probes used for quantification were corrected for nonspecific binding according to results with probe NON338 as a negative control.

Analysis of amoA sequences. To supplement results from FISH and the specific 16S rDNA library, comparative sequence analysis of biofilm-derived 491-bp fragments of the ammonia monooxygenase gene (amoA) was performed as previously described (42). Amplificates of amoA were separated according to their GC content by agarose gel retardation as described by Schmid et al. (47).

Nucleotide sequence accession number. The 16S rDNA partial sequences obtained in this study are available from the EMBL nucleotide sequence database under accession no. AJ297415 to AJ297419. The amoA partial sequences appear under accession no. AF293065 to AF293075 and AY007575.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Functional analysis. Concentrations of the solutes measured by microsensors in the bulk liquid of the measuring setup closely reflected the bulk liquid composition in the reactor as determined independently by online monitoring (data not shown).

2

1

View larger version (20K): [in this window] [in a new window]   FIG. 1.   Representative examples of vertical concentration microprofiles of oxygen, nitrate, and nitrite measured in biofilm at different times of the reactor cycle. Numbers refer to the time in minutes after start of the treatment cycle (start of aeration, t = 180 min).

Calculation of nitrate production. Volumetric nitrate production rates were separately estimated for the productive surface (depth of 0 to 200 µm) and the deeper biofilm (300 to 600 µm) from profile data measured in four different batch runs. The deeper layer showed some denitrifying activity up to about 1.0 µmol cm³ h¹ at t = 360 min, evolving together with nitrifying activity in the upper layer. The nitrate production in the surface layer during aeration reached maximum estimated rates of 1.7 µmol cm³ h¹ (corresponding to 0.03 µmol cm² h¹) and thus accounted for about 7% of the oxygen uptake at the end of the process (Fig. 2).

View larger version (18K): [in this window] [in a new window]   FIG. 2.   Volumetric net production and consumption of nitrate at the biofilm surface (0 to 200 µm []) and in the deeper biofilm (300 to 600 µm []) during the reactor cycle. Data points originate from measurements of nitrate microprofiles of four different batch runs. For details of calculations, see text.

Broad-scale community structure. Among all phylogenetic groups tested, four groups dominated the biofilm community, as shown by FISH: (i) members of the gram-positive bacteria with high DNA G+C content (GPBHGC), which were mainly coccoid cells forming loose aggregates; (ii) members of the -proteobacteria, forming dense layers at the very surface and dense globular aggregates mostly located within the upper 200 µm; (iii) a population morphologically similar to the GPBHGC, with cells typically arranged in tetrads, that hybridized with probes ALF968, ALF1b, and GAM42a, but not with probes for subgroups of the -proteobacteria (Fig. 3A), leaving their phylogenetic affiliation as yet unresolved; and (iv) members of the phylum Nitrospira (see below). Other phylogenetic groups together did not account for more than 20% of the microbial community in the biofilm (data not shown). The vast majority of all cells were located in the upper 300 µm of the biofilm, whereas in the deeper, permanently anoxic layers, only a few cell aggregates could be detected occasionally.

View larger version (45K): [in this window] [in a new window]   FIG. 3.   Confocal laser-scanning micrographs of vertical thin sections of biofilm as hybridized with different fluorescent oligonucleotide probes. (A) Overview of the biofilm surface hybridized with ALF968 and HGC1351 (in green and red, respectively). (B) Aggregates of -subclass AOB as hybridized with Nso1225 (green) and NOB of the genus Nitrospira (Ntspa662, red). (C) Clustered aggregates (arrows) of members of the N. communis lineage (NmII, red) among the population -subclass AOB (Nso1225, green). Colocalization of the two probes results in a yellow color. (D) Aggregates of members of the N. europaea lineage as hybridized with probe NEU (green) and of the N. oligotropha lineage as hybridized with probe NOLI191 (red). (E) Colocalization of signals after hybridization with probes specific for the -subclass AOB (Nso1225, green) and the N. oligotropha lineage (Nmo218, red); the insert shows a big aggregate as hybridized with probes Nso1225 (green) and NOLI191 (red). (F) Orthogonal representation of a dense assemblage of Nitrospira sp. as hybridized with probe Ntspa662. Scale bars are 50 µm (A, B, D, and E), 25 µm (C), and 5 µm (insert in panels E and F), respectively. Dashed lines indicate the surface of the biofilm exposed to the wastewater.

AOB community structure. A common problem for the quantification of nitrifying bacteria is the formation of dense aggregates resulting in a typical patchy distribution of AOB and NOB (Fig. 3). Therefore, neither normal distribution of values nor homogeneity of variances is given throughout the biofilm. For that reason, the median may be a better representative of cell densities than the mean and will be given in the following sections. To allow comparison with other studies, however, both means and medians are displayed in Fig. 4.

View larger version (24K): [in this window] [in a new window]   FIG. 4.   Depth distribution of -subclass AOB (A), of populations affiliated with different lineages of the genus Nitrosomonas (B), and of the NOB Nitrospira spp. (C) in the biofilm investigated. Numbers are given as volume-specific abundances. Cumul, cumulative. Note the different scales of the ordinates.

3

3

3

AOB-specific PCR. Because hybridization with a single probe is sometimes not sufficient to prove the identity of a given cell (2), a specific PCR and cloning strategy was applied to support the occurrence of populations affiliated with the N. oligotropha lineage within the biofilm. From the 26 clones analyzed by ARDRA, 10 different restriction patterns were obtained. Three of the patterns, representing a total of 13 clones, were indicative of sequences most similar to Nitrosomonas isolate JL21 (55), a member of the N. oligotropha lineage (Fig. 5). Two patterns, representing a total of eight clones, belonged to sequences most similar to Nitrosomonas communis Nm2. The remaining five clones with different ARDRA patterns represented single sequences not related to the genus Nitrosomonas. The amplification of these sequences and of the ones similar to Nitrosomonas communis was due to insufficient primer discrimination under the conditions chosen.

View larger version (28K): [in this window] [in a new window]   FIG. 5.   Phylogenetic tree of the genus Nitrosomonas inferred from comparative analysis of 16S rDNA sequence data. The accession numbers of the published sequences used are given in the tree. Sequences with accession no. AB000701, AB000702, AJ005546, M96399, M96402, M96403, and Z46987 represent the N. europaea lineage, and sequences with accession no. AJ003777, M96400, Z46990, Z69091, and Z69097 represent the N. marina lineage. The numbers of clones with identical ARDRA patterns for each sequence are given in parentheses. Phylogenetic reconstruction is based on a maximum-likelihood tree calculated from 950 informative positions with a genus-specific 50% positional variability filter. Tree topology was tested by distance matrix and maximum-parsimony methods, and a consensus tree was drawn. Multifurcations connect branches for which a relative order could not unambiguously be determined by the different treeing methods used. Bootstrap values (100 cycles) refer to the maximum-parsimony tree. Values smaller than 80% were omitted. The sequence of N. oligotropha isolate Nm45 was added to the consensus tree by the ARB maximum-parsimony method without changing the tree topology. The bar represents 10% estimated sequence divergence.

Design and application of probe Nmo218. The sequence of probe NOLI191 differs from the target sites of most of the recently reported 16S rRNA sequences of the N. oligotropha lineage (42, 52, 55). Furthermore, the probe hybridizes to organisms (e.g., Pseudomonas lemoignei) not belonging to the intended target group. Therefore, we designed probe Nmo218 encompassing the whole N. oligotropha lineage (Fig. 5). The probe sequence and binding sites of target and nontarget organisms are displayed in Table 1. The dissociation temperature of the probe was 40.1 ± 1.7°C, and the optimal formamide concentration in the hybridization buffer was 35%. When applied under these conditions, probe Nmo218 did not hybridize with any negative controls, except for Nitrosomonas cryotolerans and Nitrosomonas aestuarii (Table 1). (The sequences of five clones with identical binding sites to N. cryotolerans or N. aestuarii might also not be discriminated.) The occurrence of N. cryotolerans can, however, be ruled out by parallel use of probe NmIV. Hybridization of probe Nmo218 to the biofilm resulted in a picture similar to that with probe NOLI191 (Fig. 3D and E). However, simultaneous hybridization with both probes revealed a certain number of organisms only hybridizing with either of the two probes. Cells of N. aestuarii hybridized with probe Nmo218 (Table 1), but not with probe NOLI191. Consequently, cells showing this hybridization pattern might be related to N. aestuarii.

AOB diversity assessed by comparative amoA sequence analysis. Specific amplification of the amoA gene fragments from extracted biofilm DNA and subsequent separation by gel retardation resulted in three clearly visible amoA bands. All bands were excised and separately cloned and sequenced. A total of 12 amoA clones (4 from each band) were analyzed, which represented five phylogenetically distinct -subclass AOB (Fig. 6). Two clusters (representing two different bands), each containing four amoA sequences, were found within the Nitrosomonas marina-N. oligotropha lineages (which cannot clearly be distinguished by using amoA sequences) (42). The third band contained a higher diversity of amoA sequences. One amoA clone was closely related to Nitrosomonas communis. Another clone was affiliated with Nitrosococcus mobilis, which could not be detected in the biofilm by FISH. The two remaining amoA clones obtained from the third band clustered together and formed an independent lineage not closely related to any described AOB.

View larger version (16K): [in this window] [in a new window]   FIG. 6.   Phylogenetic FITCH-Margoliash amoA tree (using global rearrangement and randomized input order [7 jumbles]) showing the position of the 12 recovered biofilm sequences in relation to described -subclass AOB (42). The bar indicates 10% estimated sequence divergence. The root was determined with the amoA sequences of the -subclass ammonia-oxidizing bacteria (42). Cloned amoA sequences with amino acid similarities >99% are represented by a single clone: the number in parentheses indicates the number of clones for each representative. Clones labeled with the prefixes B2, B3, and B4 were obtained from different gel retardation bands.

NOB community structure. The only NOB detected in the system were of the genus Nitrospira, as identified by hybridization with probes Ntspa712, Ntspa662, and NSR826 (Fig. 3B and F). Low numbers of cells also hybridized with probe NSR1156. The abundance of Nitrospira spp. in the upper 100 µm of the biofilm was 1.1 × 10¹¹ cm³ and therefore was about 30 times higher than the abundance of -subclass AOB, as quantified with probe Nso1225. In comparison to the AOB, the vertical distribution was broadened towards depth. Within the section investigated, a 95% limit in cumulative (mean) abundance was reached, although not before a depth of 400 µm, and the numbers were high even in deeper layers of the biofilm (Fig. 4C).

Cell-specific nitrification rates. The average cell density of NOB in the upper 200 µm of the biofilm where nitrification was mainly performed was 6.6 × 10¹⁰ cm³. Because nitrate production in this layer during aeration on average was 0.7 µmol cm³ h¹, conversion rates were about 0.01 fmol of nitrite cell¹ h¹ for the final aeration period. Maximum rates were 0.025 fmol of nitrite cell¹ h¹. Cell-specific ammonia oxidation rates can only be roughly estimated based on nitrate production rates. With the average abundance of AOB of 1.09 × 10⁹ cm³ in the upper 200 µm, mean conversion rates were at least 0.65 fmol of ammonium cell¹ h¹ (maximum of 1.5 fmol of ammonium cell¹ h¹). However, because of the accumulation of nitrite within the nitrification zone, this number is clearly an underestimate.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Biofilm activity and community structure. In the biofilm studied, biological phosphorus removal is combined with nitrogen removal via nitrification and denitrification. Nitrification is typically performed by distinct phylogenetic groups and will be discussed later. For biological P removal, involvement of various groups has been suggested: e.g., GPBHGC (12, 59), members of the genus Rhodocyclus (7, 8, 21), or a mixture of multiple species (34). The high abundance of GPBHGC and the detection of DAPI-stainable inclusions indicative of polyphosphate (23) in part of this population strongly promote the first hypothesis for our system. However, the fixation procedure used might have led to the disappearance of polyphosphate granules (21), and therefore PAO, especially among gram-negative bacteria, might have been overlooked.

Competition for oxygen. Both nitrification and phosphate uptake require oxygen. Therefore, competition for oxygen between the respective microbial populations during the aeration period is to be expected. Microsensor data indicate that the nitrifying bacteria were oxygen limited. Oxygen penetration was low, whereas ammonium was present in excess during the whole process. The distribution of AOB and NOB in general corresponds to the oxygen penetration depth, supporting the findings of earlier studies (39, 50). The delayed onset of nitrification after the start of the aeration and the accumulation of nitrite despite the high abundance of NOB very likely reflect the limited supply of oxygen. Due to their high K_m for oxygen, AOB and NOB are poor competitors compared to heterotrophic bacteria (17, 41, 56). Thus, during the initial aeration period, oxygen is taken up preferentially by heterotrophs as the PAO. As the phosphate uptake rate declines with the ongoing aeration period (data not shown), the stoichiometric oxygen demand of PAO metabolism decreases as well (51). Consequently, during the course of the aeration period, phosphate accumulation is most likely replaced by nitrification in terms of oxygen consumption.

Nitrification rates. The handling of concentration data to estimate volumetric nitrate production rates (see Materials and Methods) is based on two assumptions: (i) low lateral heterogeneity of the concentration in the biofilm at a certain time and (ii) slow dynamics of the concentration changes due to activity compared to those due to diffusion (i.e., a pseudo-steady-state situation). Structural analysis showed a certain lateral inhomogeneity based mostly on the clustering of nitrifiers (Fig. 3B), but inhomogeneity is low in the horizontal direction, and the resolution of the measurements was not in the range of the cluster size. Therefore, the first assumption is not violated. For a similar system (36), it was shown that the time required to reach equilibrium was 1.8 min in a biofilm with a thickness of 500 µm. This is quite short compared to the overall cycle dynamics observed here. Nevertheless, the second assumption may introduce some error, especially in highly dynamic layers. The rates determined here should therefore be regarded as best estimates.

Community structure of AOB. The combined approach of FISH, 16S rDNA, and amoA analyses led to the unexpected but consistent finding that there were several different AOB populations occurring in close spatial vicinity. In contrast, previous studies in biofilms and activated sludge usually reported a single population dominating the system, e.g., populations related to N. europaea (39, 50), Nitrosococcus mobilis (22), or members of the genus Nitrosospira (39, 49). This raises the question of what mechanisms allow for the coexistence of the different AOB populations in the biofilm studied here.

3

NOB population. Using different probes for NOB, the existence of a certain population of Nitrospira spp. could be proven, whereas Nitrobacter spp. were not detected. This is in agreement with several other culture-independent studies of engineered systems (10, 22, 39, 48, 49, 61). Analysis of the probe match pattern in the current data set by the software package ARB (Technical University Munich, Munich, Germany) showed that it fits a cluster of closely related sequences in the genus Nitrospira (strains with accession no. Y14636 to Y14643) retrieved from a nitrite-oxidizing sequencing batch reactor by Burrell et al. (10). The NOB population in our system is thus likely to be affiliated with this particular cluster of the genus Nitrospira. It might be speculated whether this cluster possesses common functional features leading to a competitive advantage under the periodically changing conditions typical for sequencing batch reactors. The cell densities of Nitrospira spp. found in this study are on the order of magnitude of those reported from a purely nitrifying fluidized bed reactor (48). However, the abundance is 1 order of magnitude higher than that of the AOB. The question remains of how such a high abundance is supported, when the substrate turnover is speculated to be low (48).

Conclusions. Through the combination of microsensor measurements and molecular methods, it was possible to resolve the structure and activity of the nitrifying community in a complex, P-removing biofilm. Some first insights were obtained into the mechanism that allows for the coexistence of different populations of AOB in the same biofilm, i.e., separation of their distributions in space and separation of their activities in time. The design of probe Nmo218 specific for the N. oligotropha lineage will enable in situ detection and quantification of N. oligotropha-like AOB in future studies to collect more information about their natural abundance and ecology.