INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Activated of sludge is currently the most widely used process for the treatment of both domestic and industrial wastewaters and, at least by scale, one of the most important microbiological technologies (18). It relies primarily on the degradation and uptake of organic matter by a microbial community under oxic conditions. Processes at modern plants are often supplemented with anoxic reactor stages to enhance nitrogen and phosphorous removal. The biomass is finally separated from the purified water by gravitational settling prior to the recirculation of part of this sludge back into the aeration basin. The process, therefore, selects for microorganisms that remain in the system due to their growth in flocs.

This immobilized growth leads to conditions that markedly differ from conditions of suspended growth in the bulk water phase. Closer interactions of different physiological types of microorganisms, e.g., of ammonia and nitrite oxidizers (22), are possible, and bacteria are better protected from protozoan grazing (13) or harmful substances. On the other hand, the transport of solutes (e.g., oxygen and nutrients) in flocs is expected to be mainly diffusional (50, 51). Although the vast majority of activated sludge flocs have been reported to be smaller than 20 µm in diameter, i.e., of a size at which diffusion limitation is unlikely, flocs larger than 50 µm contribute the most to surface area, volume, and mass (28). In these larger flocs, the development of anoxic zones has been postulated due to diffusion limitation (see e.g., reference 50). This possibility allows for combined nitrification-denitrification in quasistratified flocs; hence, reaction space and time would be saved (16 and 51 and references therein). Less beneficial, anoxic microniches may also support the survival and activity of sulfate-reducing bacteria (SRB) in aerated activated sludge, resulting in the production of H₂S and subsequent problems with sludge bulking (58) or floc disintegration (38).

These hypotheses have been supported indirectly by several reports of nitrogen losses from aeration basins (e.g., in references 16, 20, and 51) and by the detection of SRB in activated sludge by cultivation (26, 58) and fluorescence in situ hybridization (FISH) (34). In contrast, no anoxic zones could be detected with oxygen microsensors in large activated sludge flocs (diameter, 1.6 mm) at air saturation (26).

Recently, a flow system was developed for microelectrode measurements in freely sinking aggregates ("marine snow") that also enables the analysis of smaller and more fragile flocs in a natural flow field (40, 41). We used this setup for microsensor measurements of oxygen, nitrate, nitrite, and hydrogen sulfide in individual activated sludge flocs. These single floc measurements were complemented with ¹⁵NO₃ and ³⁵SO₄² incubation experiments (15, 37) to determine overall rates of denitrification and sulfate reduction in the different sludges tested. Finally, the three-dimensional structure, which is critical for the transport mechanism in a floc (diffusion or advection), was recorded by confocal laser scanning microscopy (CLSM) and the samples were screened for SRB by FISH with rRNA-targeted oligonucleotide probes (4, 34) and by PCR with primers specific for the dissimilatory sulfite reductase gene (53). By this multiple-method approach we hoped to achieve a more comprehensive picture of the occurrence and preconditions of anaerobic processes in activated sludge flocs.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Samples. Activated sludge samples were obtained from the aeration basins of municipal wastewater treatment plants (WWTP) in Bremen-Seehausen (Germany), Aarhus-Marselisborg, Odder (both Denmark), and Prague (Czech Republic), and from two lab-scale sequencing batch reactors (SBR) receiving artificial wastewater (peptone, 1,000 mg chemical oxygen demand (COD) liter¹; acetic acid, 300 mg COD liter¹; glucose, 400 mg COD liter¹; ethanol, 300 mg COD liter¹, total N, 82 mg liter¹; N-NH₄⁺, 0.2 mg liter¹; total P, 14 mg liter¹). The sulfate concentration was 102 mg of SO₄² liter¹. Both SBR were operated with rapid filling periods (5 to 10 min) to simulate the conditions in a plug-flow reactor with high-substrate-concentration gradients. SBR1 was operated with a complete oxic cycle (23 h of aeration, 1 h of settling), whereas SBR2 was subjected to an alternating cycle (3 h of anoxic conditions, 8 h of aeration, 1 h of settling). Some operational data of the investigated sludges are summarized in Table 1.

Microsensor measurements. Clark-type microsensors for O₂ (42), LIX-type microsensors for NO₂ and NO₃ (8), and amperometric H₂S microsensors (25) were constructed, calibrated, and used for measurements as previously described. The lower detection limits of NO₂ and H₂S were 0.1 and 1 µM, respectively. Microprofiles of single activated sludge flocs were recorded by keeping the flocs freely suspended in a vertical-flow system, where the flow velocity opposed and balanced the sinking velocity of the individual floc. To create a parallel, nonturbulent, uniform flow, a nylon stocking was mounted in the flow chamber horizontally to the flow and the flocs were positioned just above this net (40, 41). With this flow system the flocs could be stabilized in the upwardly flowing water column and microsensor measurements with a spatial resolution of 25 to 50 µm were possible from above, i.e., downstream of the floc, without disturbing the flow field (40). For practical reasons, microprofiles of different chemical species were usually recorded in different flocs.

1

Calculations. The volumetric oxygen respiration rate (R) of a sphere with zero-order kinetics at steady state is described by the following equation (41):

(1)

where r₀ is the radius of the sphere, 4r₀² and 4/3r₀³ are the surface area and volume, respectively, r_c is the radial distance from the center at which the oxygen concentration becomes 0 (if there is no anoxic zone, r_c equals 0), D_W(ox) is the molecular diffusion coefficient of oxygen in water, C and C₀ are the concentrations of oxygen in the bulk water phase and at the floc surface, respectively, and _eff is the effective thickness of the diffusive boundary layer (DBL) (see Fig. 1). The same formula was used to calculate nitrate reduction rates of single flocs from nitrate microprofiles. The D_W for oxygen at 20°C is 2.12 · 10⁵ cm² s¹ (5), and that for nitrate is 1.66 · 10⁵ cm² s¹ (29). Determination of _eff and data processing were done by a simple diffusion-reaction model by assuming zero-order kinetics as described in detail by Ploug et al. (41).

(2)

5

1

¹⁵NO₃ incubations. Denitrification rates were determined with a batch reactor with a liquid volume of 1.2 liters and a gas volume of 0.5 liter (including tubing). The reactor was cylindrical with a diameter of 10 cm and a height of 17 cm. The bottom section was funnel shaped with a porous glass grid in the center, through which gas was supplied. This arrangement prevented the development of stagnant zones. The gas flow rate was just sufficient to keep the flocs in suspension. Oxygen concentration measurements at different positions within the reactor showed that it was well mixed under test conditions. Prior to the incubations, N₂ in the reactor was exchanged by argon to lower the background, thereby improving the detection of ¹⁵N-enriched N₂. Rate measurements were performed at air saturation, at an oxygen concentration of 40 to 60 µM, and in the absence of oxygen by adjusting the oxygen/argon ratio in the gas supply. An oxygen microelectrode was inserted in the reactor for continuous monitoring during the experiments. Three hundred milliliters of concentrated activated sludge was added to the reactor, which was then filled up with 1 liter of artificial wastewater (as described for the microsensor measurements) and amended with sodium acetate to a concentration of 7.8 mM. The reactor contained 2 to 4 g of total suspended solids (TSS) liter¹. After the oxygen concentration was adjusted, 8.3 ml of Na¹⁵NO₃ was added from a 12 mM stock solution of 99.2 atom% of ¹⁵NO₃, corresponding to a final concentration of 100 µM ¹⁵NO₃. During the 30-min incubation experiment, gas samples of 1 ml were taken from the reactor headspace every three minutes through a septum with a gas-tight syringe (model 1001RN; Hamilton) and transferred to gas-tight exetainers (Labco) that had been filled with N₂-free distilled water.

³⁵SO₄² incubations. Sulfate reduction rates were determined by the ³⁵S-radiotracer method (15) in samples from SBR1, SBR2, and the WWTP at Prague. Reactor design, incubation conditions, and filling of the reactor were as described for the ¹⁵N experiments. After the oxygen concentration was adjusted, 20 ml of tracer was added (Na₂³⁵SO₄, 2 MBq). Through a septum, samples of 5 ml were taken from the reactor during the first 10 minutes once per minute and then for another 10 minutes once every two minutes. Subsequently, samples were taken every 10 minutes until 1 h after the start of the test. To each sample, 5 ml of fixation solution (20% Zn acetate, 1% formaline [pH 5]) was added, and the mixture was shaken well. Samples to which 0.1 ml of tracer was added after fixation were used as blanks for each incubation. Fixed samples were stored at 4°C until further analysis within 2 months. The samples were then centrifuged at 10,000 × g, and the reduced sulfur species in the pellet were determined by single-step chromate distillation according to the method of Fossing and Jørgensen (15). The detection limit of the method was a sulfate reduction rate of 5 µmol of S g of TSS¹ h¹.

SRB screening. Activated sludge samples were fixed with paraformaldehyde, immobilized on microscopic slides, and dehydrated as described previously (3). For FISH, a set of oligonucleotide probes specific for SRB was used, specifically, probe SRB385 (2), which targets a broad range of SRB but also binds to numerous non-SRB (e.g., myxobacteria, clostridiae, and actinomycetes [34]); probes DSV698, DSV407, DSV1292, and DSV214, specific for the family Desulfovibrionaceae (34), probes 221 and 660, specific for the genera Desulfobacterium and Desulfobulbus, respectively (10); and probes DSB985 and DSS658, specific for the genus Desulfobacter and the taxon Desulfosarcina-Desulfococcus (34), respectively. All probes were purchased labeled with the fluorescent dye CY3 (Interactiva Biotechnologie, Ulm, Germany) and used in FISH by using the protocol and the conditions recently described by Manz et al. (34). After the hybridization procedure the samples were stained with 4',6-diamidino-2-phenylindole (DAPI) according to the method of Wagner et al. (52), mounted with antifading reagent (Vectashield; Vector Laboratories Inc., Burlingame, Calif.), and examined under an epifluorescence microscope (Carl Zeiss, Jena, Germany).

Three-dimensional (3D) analysis of flocs. For staining with fluorescein isothiocyanate (FITC), which covalently binds to proteins (19), 0.2 ml of settled flocs was added to 15 ml of staining solution (0.1 M sodium phosphate [pH 7.0], 4 mg of FITC liter¹). After 5 min of gentle mixing, the aggregates were allowed to settle, the solution was decanted, and the flocs were washed twice with 0.1 M sodium phosphate, pH 7.0. The aggregates were stored at 4°C in 0.1 M sodium phosphate (pH 7.0) with 4% paraformaldehyde. For CLSM analysis, the pH was raised to 9 by the addition of 1 M carbonate buffer. Staining with calcofluor to visualize polysaccharides was performed similarly in the same buffer with 300 mg of calcofluor liter¹ (7). The staining time was 2 h. Washing and storage were as described above. The aggregates were microscopically examined at pH 7.0. DNAs within the flocs were stained with ethidium bromide (1 µg ml¹) for 15 min in the same buffer. The flocs were washed as described above and immediately analyzed by CLSM.

CLSM analysis. FITC-, calcofluor-, and ethidium bromide-stained flocs were transferred in 1 ml of buffer to a chamber sealed on the bottom with a glass cover and analyzed with an inverse CLSM (model LSM510; Carl Zeiss). A 40× objective (model Plan-Neofluar 1.3) was used and three different lasers (UV wavelength, 351 plus 364 nm; Ar ion wavelength 458 plus 488 nm; HeNe wavelength, 543 nm) were applied for excitation. Image processing, including 3D reconstruction, was performed with the standard software package delivered with the instrument (version 2.01, service pack 2). Images were printed on a Kodak printer 8650 by use of the software package Power Point (version 7.0; Microsoft).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Microprofiles. We took microsensor measurements for the different parameters in 250 individual activated sludge flocs with a size range of 400 to 2,300 µm (maximum length as observed by dissection microscopy). Larger flocs often consisted of a loose agglomeration of compact subunits of 50 to 100 µm, suggesting a dynamic aggregation and disintegration. Flocs smaller than 400 µm could not be sufficiently stabilized in the flow chamber for profiling.

3

1

3

1

3

1

3

1

3

1

View larger version (34K): [in this window] [in a new window]   FIG. 1.   Typical microprofiles of oxygen, nitrite, and nitrate in activated sludge flocs from the WWTP at Prague (A) and SBR1 (B), measured in the net-jet flow chamber with 2 mg of O2 liter1. The dotted nitrate profile in panel A was recorded under anoxic conditions and displays the nitrate reduction potential of the floc. All profiles shown were measured in separate flocs of the same size, and the data were compiled for the two panels displayed. Shaded area, floc; r0, floc surface; rc, distance from the floc center where oxygen disappears; eff, effective DBL.

3

1

Respiration rate. Volumetric respiration rates (R) of individual flocs were calculated from the oxygen profiles by assuming diffusion to be the only transport process. They were between 0 and 18 nmol of O₂ mm³ h¹, with the highest R values being found in those sludges in which anoxic microniches had been detected, i.e., in SBR1, SBR2, and the WWTP at Bremen (Fig. 2). Under 40 to 60 µM oxygen, the respiration rates obviously decreased with floc size (Fig. 3B), while under air saturation this trend was somewhat less pronounced (Fig. 3A).

View larger version (23K): [in this window] [in a new window]   FIG. 2.   Mean volumetric respiration rates (R) and mean floc sizes (d) with standard deviations (T bars) of all flocs from all samples in which oxygen gradients were measured. The dotted line indicates a floc diameter of 1 mm. *, sludges in which anoxic zones were detected.

View larger version (19K): [in this window] [in a new window]   FIG. 3.   Volumetric oxygen respiration rates versus floc diameter as measured under air saturation (A) and under 2 mg of O2 liter1 (B). Arrows indicate flocs where the oxygen profile was most likely influenced by advective transport (liquid flow). These data points were excluded from the regression curve, as the respiration rates are most probably underestimates (see Discussion).

Incubation experiments. Batch experiments to determine denitrification and sulfate reduction rates by stable isotope and radioisotope techniques were performed under the same conditions and with results similar to those of the microsensor measurements (Table 2). Under air saturation virtually no denitrification occurred, and under reactor conditions the rates were extremely low except with the SBR samples. All sludges were, however, capable of denitrification under anoxic conditions. Sulfate reduction could not be detected either in sludge from the Prague WWTP or in sludges from the SBR, regardless of the aeration conditions applied (air saturation, reactor conditions, or anoxic conditions).

SRB screening. FISH with probe SRB385 suggested the presence of SRB in all tested sludges. The abundance of specifically hybridized cells was roughly estimated to be 1 to 2% in the SBR and 3 to 5% of total cells stained by DAPI in all other samples. Of the more specific probes, only DSV698 and DSV1292, complementary to the majority of Desulfovibrio species, detected significant numbers of target cells, i.e., 0.5 to 1% in the SBR and 2 to 4% of total cells in the other samples. In comparison, members of the genera Desulfobacterium, Desulfobacter, Desulfobulbus, Desulfomicrobium, and Desulfosarcina detected by FISH together made up less than 0.2% of total cells in all samples. Additionally, DNAs were extracted from activated sludge samples of the SBR as well as from samples of the WWTP at Prague and Bremen. Using the same amount of DNA (ca. 20 ng) for the PCR, we obtained no PCR product of the DSR gene fragments from the SBR samples but we retrieved distinct PCR products of the expected size from the WWTP samples (data not shown). As DSR is a key enzyme for sulfate reduction, the detection of its genes indicates the presence of SRB (or of at least their DNAs) in the WWTP but not in the SBR activated sludges.

3D analysis of activated sludge flocs. For a qualitative analysis of the 3D floc structures of different sludges (WWTP at Prague and Bremen, SBR1, and SBR2), flocs were stained with either FITC, calcofluor, or ethidium bromide. These fluorescent dyes bind to proteins, polysaccharides, and DNA, respectively, which represent the main compounds of the extracellular polymeric substance of activated sludge flocs (50). Staining of these substances should therefore result in the visualization of the floc's structure. A comparison of the different stains revealed that, at the low resolution needed to visualize complete flocs, all three dyes yielded approximately the same picture, i.e., the same ratio of stained floc material to unstained pore volume. However, as FITC-conferred fluorescence of the flocs was brightest and gave the best CLSM images, all further 3D analyses were performed with FITC-stained flocs.

View larger version (45K): [in this window] [in a new window]   FIG. 4.   CLSM images (for the red-green overlay, use red-green glasses) of the 3D structures of activated sludge flocs from the WWTP at Prague (A) and SBR1 (B) after FITC staining. Size of pictures, 300 by 300 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Anoxic microniches and denitrification. When incubated under air saturation, anoxia was never detected inside activated sludge flocs, which is in agreement with the measurements of Lens et al. (26). Calculation of acetate concentrations in the centers of the flocs, based on the measured oxygen respiration rates (equation 2), showed that in only 2 of 35 flocs could acetate be completely depleted. Therefore, assuming that acetate (this study) or glucose and starch (26) were suitable substrates for the microbial community in activated sludge, respiration was most likely not limited by the availability of organic carbon. These results indicate that the respiration capacity of activated sludge is simply not sufficient to create anoxia under air saturation. However, even when flocs were incubated under more realistic conditions (2 mg of O₂ liter¹), no anoxic zones and no nitrate reduction were detected except in activated sludge flocs of the SBR and a few flocs from the Bremen WWTP. The question of how representative the microsensor measurements from single flocs were, i.e., if the measuring approach was suited to detect anoxic microniches and if the results are meaningful for a complete activated sludge basin, has to be raised. In our study, we analyzed only flocs larger than 400 µm. Although this size class is present in activated sludge only in low numbers compared to the numbers of smaller flocs, flocs of this size contribute most to the volume and mass of the sludge (12, 28) and hence are most important for the activity of a plant. Furthermore, anoxic zones due to diffusion limitation are primarily to be expected in larger flocs since the volumetric respiration rates required to create anoxia exactly at the center of a floc increase with the square of the floc radius (equation 2) (Fig. 5) (41). Therefore, under reactor conditions, volumetric respiration rates of more than 70 nmol mm³ h¹ are necessary for anoxia in flocs with a diameter of 400 µm or less. Such high rates have been found only in microbial mats (21) and nitrifying aggregates (9), while the rates reported from various other systems such as detritus pellets (41), trickling filter biofilms (23, 24), microbial mats (14), and sediments (14, 43) are all in the same range (1.2 to 39.6 nmol mm³ h¹) as the values measured in this study (0 to 19 nmol mm³ h¹). It is thus questionable if the respiration rates required for anoxia in flocs smaller than 400 µm can ever be reached in activated sludge (Fig. 5). Furthermore, we measured microprofiles by simulating sinking flocs, i.e., in a flow chamber with laminar flow, where no turbulent mixing or collisions of flocs occur. The latter processes, however, are typical of aerated activated sludge basins and obviously lead to a steady aggregation and disintegration of flocs. Gradients and flocs are consequently dynamic features; e.g., the center of a floc might become exposed to oxygen again after an anoxic period by the disruption of the floc. For these two reasons, the sizes of the studied flocs and the measuring conditions, it was more likely to overestimate anaerobic processes in the activated sludges by microsensor analysis rather than to overlook them.

View larger version (24K): [in this window] [in a new window]   FIG. 5.   Volumetric oxygen respiration rates required to create anoxic conditions exactly at the center of a floc with a maximum DBL under different bulk oxygen concentrations (calculated from equation 2). Under such diffusion-limited conditions, R decreases with the square of the floc radius. The shaded area represents respiration rates and floc diameters that have been measured during this study. The dotted line represents the highest volumetric respiration rates measured in microbial communities (9, 21).

1

1

1

1

1

Respiration rates. Valuable information can be derived from the measured volumetric respiration rates (R) of individual flocs in relation to their size. As displayed in Fig. 5, smaller flocs require higher respiration rates to become anoxic because of their higher surface to volume ratios. This also applies to the investigated samples (Fig. 2). For example, the mean floc sizes of analyzed flocs from SBR1 and the Aarhus WWTP are almost the same; however, only the respiration rates of SBR1 are sufficient to create anoxic zones. Furthermore, the volumetric respiration rate was negatively correlated with floc size (d) (Fig. 3), which was somewhat more pronounced for flocs under reactor conditions than for flocs under air saturation. A decrease of R with d² usually indicates diffusion limitation within the floc (Fig. 5). The observed decrease of R with d^1.5 at 40 to 60 µM oxygen (Fig. 3B) might be explained by the different extents of oxygen limitation in the different flocs that contributed to the regression analysis since (i) some flocs are truly diffusion limited and have anoxic centers, (ii) some flocs might slow down respiration rates due to rather low oxygen concentrations (1 to 5 µM) in their centers, and (iii) some flocs with higher oxygen levels in their centers respire with maximum rates since the K_m for oxygen consumption by heterotrophic bacteria is about 1 µM (56). In contrast, diffusion limitation can be excluded from the reasons for this trend under air saturation (Fig. 3A). Oxygen as well as organic carbon is present in concentrations high enough to prevent any limitations, as has been discussed before. An alternative explanation lies in the structure and geometry of the floc. Larger flocs might be less dense than smaller ones, i.e., contain fewer active cells per volume and more voids and dead material. This seems to be partially the case, as large flocs (>1 mm in diameter) often enclose bigger particles which do not contribute to respiration. Furthermore, activated sludge flocs are fractal in their geometry, with fractal dimensions of 1.0 to 1.8 (27, 33, 48). As is typical of fractal aggregates, the porosity of flocs increases with increasing floc size (1), resulting in reduced mass and hence respiration rate per volume as the aggregates get larger. The same also of course applies under the lower oxygen concentrations. Fractal geometry of activated sludge flocs might therefore provide an explanation of why the volumetric respiration rates of most sludges have been too low to create anoxic conditions even in the larger flocs. Alternatively, the lack of anoxic microniches might be due to advective transport through pores and channel-like structures (32) since larger flocs often consist of dense subunits that are only loosely connected. Flow might have been detected in several oxygen profiles that showed a local increase in oxygen concentration (Fig. 3). In the absence of photosynthesis, advective transport of oxygen-rich bulk medium through pores into the floc is the only process that could explain this observation. Flow can substantially enhance oxygen transfer compared to diffusion. In this case, our calculations of volumetric respiration rates (and denitrification rates) based on diffusional transport underestimate R. If advection is an important transport mechanism in activated sludge flocs, reduced oxygen bulk concentrations do not necessarily result in anoxic zones and enhanced denitrification as discussed above.

Sulfate reduction. Low numbers of SRB were detected by FISH in all samples, and the amplification of the DSR gene fragments indicated the presence of SRB at least in samples from the WWTP at Bremen and Prague. In contrast, no sulfate reduction was detected in any experiment by microsensor or radioisotope analysis. The sensitivities of the applied techniques were relatively high (1 µM H₂S and 5 µmol of S g of TSS¹ h¹ for microsensor and radioisotope analyses, respectively) and immediate reoxidation by oxygen or nitrate was excluded during anoxic incubations in nitrate-free medium. By the ³⁵S analysis even the immediate precipitation of H₂S, e.g., as FeS (38), would have been detected. However, a shortcoming of the approach was the use of acetate as the sole carbon source in almost all experiments. Acetate is not utilized as an electron donor by incompletely oxidizing SRB, including Desulfovibrio sp. (55), which was detected as the main component of the SRB community in the analyzed samples. Sulfate reduction in our experiments had therefore to rely on endogenous electron donors of the activated sludge that were either produced from the acetate added or had been stored within the flocs. Whereas this assumption is doubtful for the microsensor experiments, due to the small volume of a single floc compared to the incubation volume and time in the flow system, it is sound for the ³⁵SO₄² incubations. In the batch reactor, about one-third of the reactor volume consisted of concentrated activated sludge and only two-thirds of the sludge bulk water had been replaced by the artificial medium. Essentially, the substrate spectrum was similar to that of the original activated sludge sample but was slightly diluted. Furthermore, no sulfate reduction could be detected after anoxic incubation of sludge with a mixture of acetate, propionate, and butyrate in a test tube. For these reasons, we presume that we were able to detect sulfate reduction in the observed systems if it had occurred. However, some of the H₂S microsensor measurements and some of the ³⁵SO₄² incubations should be repeated with more complex substrates to eventually prove our results and assumptions.

1