biofilm development in laboratory methanogenic fluidized bed reactors

Biofilm Development in Laboratory Methanogenic Fluidized Bed Reactors

L. G. M. Gorris, J. M. A. van Deursen, C. van der Drift, and G. D. Vogels’ Department of Microbiology, Faculty of Science, University of Nijmegen, Toernooiveld, NL-6525 ED Nijmegen, The Netherlands

Accepted for publication March 7, 7988

Biofilm development on sand with different heteroge- neous inocula was studied in laboratory-scale methanogenic fluidized bed reactors. Both the course of biofilm formation during reactor start-up and the bacterial composition of newly developed biofilms at steady-state were found to be similar irrespective of the type of inoculum applied. Biofilm formation proceeded according to a fixed pattern that could be subdivided in three consecutive phases, designated as the lag phase, biofilm production phase, and steady-state phase. Methano- genic activity and biomass content of the fluidized bed granules were found to be accurate parameters of the course of biofilm formation. More indirect parameters monitored did not give unambiguous results in all instances. The composition of the newly developed biomass as assessed on the basis of potential methanogenic activities on different substrates and of the concentration of specific methanogenic cofactors was consistent with electron microscopic observations.

INTRODUCTION

Anaerobic purification of the soluble organic fraction of industrial waste waters can be accomplished successfully by use of biological treatment systems such as the fluidized bed (FB) reactor. In this system, retention of purify- ing bacteria is achieved by immobilization on a mobile carrier. When sand with a particle diameter of 0.2- 0.5 mm is used as the carrier material,”2 a surface area of over 2000 m2/m3 is available for microbial growth and biomass concentrations of 30-40 kg VSS (volatile sus- pended solids)/m3 can be obtained. In general thin biofilms are formed in FB-reactors as a result of several factors, e.g., shear stress, nutrient concentration, and surface area. The presence of a thin biofilm minimizes diffu- sional limitations. The sand grains covered with biomass are maintained in a fluidized state through the upward flow of the waste water, which results in good mixing and de- gassing. The settling velocity of the FB-granules may be up to 50 m/h, allowing a high flow rate to be applied without particle carryover in the effluent. Due to the high

* To whom all correspondence should be addressed.

Biotechnology and Bioengineering, Vol. 33, Pp. 687-693 (1989) 0 1989 John Wiley & Sons, Inc.

flow rate, waste water sediments wash through the reactor and a decrease in sludge activity is avoided.

At Gist Brocades (Delft, The Netherlands), the FB-system is already used at full industrial scale in a 2-stage process for the anaerobic treatment of waste waters originating from yeast and penicillin prod~ction.~ However, a better understanding of the microbial basis of biofilm development could be exploited to improve process performance and control.2 Factors which influence the microbial population dynamics during start-up and steady-state operation of anaerobic FB-reactors4 and of other retained biomass systems59627 are currently under investigation. In order to study these factors in methanogenic FB-reactors with sand as the carrier material, a laboratory scale set-up was designed and used to investigate the influence of different microbial inocula on biofilm development.

MATERIALS AND METHODS

Experimental Conditions

The experiments were performed with six all-glass FJ3- reactors, designated as R1 to R6, in an experimental set-up described in Figure 1. The effective part of each reactor consisted of a glass cylinder with a conical bottom and a water jacket. A settler and biogas collection compartment was constructed from a wider glass cylinder and an in- verted funnel. It was equipped with a gas outlet connected to a Mariotte flask. Influent liquid was pumped into the reactor via a hook-shaped inlet tube. It was composed of synthetic waste water, liquid from the settler compartment and, in the cases specified below, of inoculum. Glass- beads, 5 mm in diameter, were used to break the force of the influent and to disperse it evenly. A more detailed de- scription of the various reactors is given in Table I. At the start, the sludge bed of each reactor consisted of bare sand with a particle diameter of 0.1-0.3 rnm and a density of 2.6 g ~ m - ~ . The speed of the recirculation pump was al- ways carefully controlled to keep the sludge bed within the effective part of the reactor.

CCC 0006-3592/89/060687-07$04.00

Figure 1. Scheme of the experimental set-up. (a) effective part of the FB-reactor; (b) sludge bed; (c) settler and biogas collection compartment; (d) biogas outlet; (e) Mariotte flask; (0 temperature bath circulator (37°C); (g) influent inlet; (h) glassbeads; (i) concentrated solution of synthetic waste water (at 4°C); (j) tap water reservoir; (k) inoculum; (1) effluent outlet equipped with water seal; (m) recirculation pump.

Table I. tions.

Specification of reactor configurations and operating condi-

Reactor number

R1 R2 R3 R4 R5 R6

Reactorvolume (mL) 900 900 650 675 650 500 Effective part (mL) 460 460 260 265 260 200 Height/diameter 6.4 6.5 8.4 8.5 8.0 9.3 HRT (h)" 1.6 1.7 1.3 1.3 1.7 2.2

(m/hIb 8.7 8.7 11.2 11.2 7.7 12.0 Bare sand (mL) 200 200 loo 100 75 75

a HRT = hydraulic retention time; determined over the total reactor volume minus the sand volume.

Vb&, = the superficial liquid velocity through the effective part.

Loading Regimen and Waste Water Composition

The organic load during start-up was adapted to the fatty acid conversion capacity of a reactor by employing the following loading regimen: ab initio the reactors received VFA (volatile fatty acids) at 0.5 g COD/h (-15 g COD/L/d), while the loading rate was doubled when the total VFA-degradation exceeded 60%; an experiment was terminated when steady-state was reached at 2.0 g COD/h. In this way, reactor overloading and substrate limitation were avoided. Thus the rate of bacterial proliferation and colonization of the support particles were reflected by the rate of biogas production.6

The reactors were fed an artificially prepared waste water containing (at 1 g COD/h): 8.4mM acetate, 2.3mM propionate and 1.9mM butyrate (3 : 1 : 1 w/v) as carbon sources; KH,PO,, K,HP04, K,SO, and NH,Cl, 0.15 g/L each; vitamins and minerals stock solutions,' 3.3 and 6.5 mL/L, respectively. A concentrated solution of the

synthetic waste water was kept at 4°C and was diluted continuously with tap water. The pH of the concentrated solution was adjusted to pH 7.0 with KOH and NaOH (molar ratio Kt :Nat = 1 : 2).

Inoculation Procedure

R1 was inoculated by the continuous addition (300 mL/h) of effluent (methanogenic activity 10 mL CH,/L/d) from a five liter FB-reactor. R2 received effluent (80 rnL/h; 4 mL CH,/L/d) from a one liter up-flow anaerobic sludge blanket (UASB) reactor. Both seed reactors were fed the synthetic waste water at 2 g COD/h. VFA and VSS contents of both effluents were below detec- tion limits. R3 and R4 were inoculated with digested sewage sludge, which had a methanogenic activity of 50 mL CH,/g VSS/d. In the case of R3, this sludge had been preactivated by anaerobic incubation for three days at 37°C in activation medium containing 4.1mM acetate, 3.3mM propionate, 2.8mM butyrate, salts, minerals, and vitamins (pH 7.0). The preactivated sludge (0.4 g VSS/L) was pumped continuously into the influent flow at 20 mg VSS/h. R4 received the same amount of inoculum which had not been preactivated at 37"C, but had been kept at 4°C after dilution in activation medium. R5 was inoculated batch-wise by addition of 15 mL mature granules (2.1 g VSS in total) taken from the above mentioned 5-liter FB- reactor. The methanogenic activity of the inoculum was 300 mL CH,/g VSS/d. R6 was inoculated both batch- wise by addition of 15 mL mature FB-granules (2.1 g VSS; 300 mL CH,/g VSS/d) and continuously by addition of preactivated digested sewage sludge (90 mg VSS/h).

688 BIOTECHNOLOGY AND BIOENGINEERING, VOL. 33, FEBRUARY 1989

Measurements and Analyses

The biogas production rate was determined by means of tap water replacement in 10-liter calibrated Mariotte flasks. The amount of methane in the biogas and the amounts of volatile fatty acids in the influent and effluent liquid were quantified by gas' and gas-liquid" chromatography, respectively. The amount of VSS per amount of ashweight (g VSS/g Aw) was determined using standard methods" as an indication of the amount of biomass (= VSS) immo- bilized on the sand (= ash). The volume of the total sludge bed was measured regularly after settling for 5 min. When stratification in the physical appearance of the sludge bed (e.g., color or diameter of sludge granules) was observed, the volume of every visually distinct homogeneous layer was measured. Samples were prepared for scanning electron microscopic examination as described elsewhere.I2

Two different methanogenic activity tests were performed. The first was employed during reactor start-up in order to assess the amount of methanogenic biomass on sand particles. For this purpose, sludge samples (0.2- 0.5 g of wet weight) were taken at regular time intervals from the middle of every homogeneous layer in the sludge bed. Fresh samples were incubated (100% N2, 37°C) in 100 mL serum bottles with 30 mL test medium, containing an excess of acetate, propionate and butyrate (1 : 1 : 1 w/v, 90 mg COD), and salts, minerals, and vitamins in the same relative amounts as in the synthetic waste water. The maximum methane production rate (pmol CH,/h) was measured over the first 4-6 d when the samples contained less than 10 mg VSS g Aw, and over the first 6-8 h of incubation at higher biomass contents. Methane production rate and biomass content were used to calculate the methanogenic capacity (pmol CH4/g Aw/h), which gives an indication of the amount of methanogenic biomass im- mobilized on the sand. The second type of activity test was performed at the end of a sludge growth experiment. Sam- ples taken from the top layer of the sludge bed were incubated with each of the following substrates: acetate (21 mg COD), butyrate (39 mg COD), propionate (3 1 mg COD) and H2/C02 (80:20 v/v, 200 kPa of pressure). The maximum methane production rate recorded in the test was used to calculate the potential methanogenic activity (pmol CH,/g VSS/h) on each substrate. The potential methanogenic activities are indicative of the relative proportions of different trophic groups of bacteria within the total biomass.'3.'4 The ratio of all four activities of one sludge will be referred to as the relative substrate spectrum.

The concentrations of specific methanogenic coractors in the biomass of the FB-granules were determined using cofactor assays based on high-performance liquid chromatography. 1 5 . l 6 The relative proportions (% of total biomass) of hydrogenotrophic and acetotrophic methanogenic biomass were derived from these values using cofactor concentrations measured in pure cultures of the methanogens indicated below as reference (per g VSS): 1.4 pmol coenzyme F420 and 37.6 pmol methanopterin in Methanobacterium formicicum grown on H,/C02; 3.5 pmol vitamin BIZ-HBI

(hbi) and 80.9 pmol sarcinapterin (spt) in acetate-grown Methanosarcina barkeri; 3.5 X pmol hbi and 2.1 pmol spt in Methanothrix soehngenii grown on acetate. The proportions of Methanothrix spp. and Methanosarcina spp. were estimated separately from the ratios of the spt and hbi concentrations. l7

RESULTS AND DISCUSSION

Course of Biofilm Development

Two types of parameters were monitored to assess the course of biofilm formation on sand during reactor start- up. Indirect parameters, viz methane production rate, total fatty acid conversion rate and volume of the total sludge bed, were measured as overall indications of biofilm formation and reactor performance. Methanogenic capacity, amount of biomass on sand and volume of individual layers within the sludge bed were determined as more direct parameters.

The results obtained with regard to the indirect parameters in the experiments with R1, R2, R3, and R5 are il- lustrated in Figure 2, while the data obtained for the direct parameters during start-up of R1 and R5 are depicted in Figure 3.

In the case of R1, start-up proceeded in a sigmoid fash- ion. An initial period of slow increases in both indirect and direct parameters was followed by a period of accelerated reactor performance. At this stage three differently struc- tured homogeneous layers, namely a top, middle, and bottom layer, could be distinguished visually in the sludge bed (Fig. 3) The bottom layer contained granules with a low methanogenic capacity and a low biomass content. These parameters remained at a low level on the bottom layer throughout the experiment. Samples taken from the middle and top layer were characterized by relatively high and increasing methanogenic activities and biomass contents. Thus, the accelerated reactor performance reflected an increase in the amount of methanogenic biomass on the sand. Eventually a plateau was reached in the various parameters.

A similar course of biofilm formation was observed with all other reactors, although the course of start-up assessed by the indirect parameters was somewhat different for R5 and R6. These parameters indicated an instant start-up (Fig. 2). Both reactors, however, had been inoculated with mature FB-granules, which formed a separate layer above the sludge bed. The volume of this inoculum layer and the methanogenic capacity of the granules present in it, both tripled within the first five weeks of operation (data not shown). In contrast, methanogenic capacity and biomass content of granules in the sludge bed remained at a low level (Fig. 3). The volume of the sludge bed did not change. The apparent fast start-up thus was due to rapid growth of organisms in the inoculum and not to substantial immobilization on sand particles other than those of the initial inoculum.

GORRIS ET AL: BlOFlLM DEVELOPMENT IN METHANOGENIC FLUIDIZED BED REACTORS 689

0 20 LO 60

2

0 20 ' LO 60 80 100 120 1LO time [days)

450 R 2

300 - 3_ - R5

c 4

Wzoo 100 0 LO 80 120 160

time (days)

Figure 2. Course of biogas production rate and volume of the total sludge bed during start-up of four FB-reactors. Arrows indicate doubling of the organic loading rate. Broken arrows indicate an accidental pH- shock

I 0.8 w - - 0

0.L -

- - . . 0 ' , , $ $ ! , ,

0 22 32 5L 6 B 83 91 165 sample day

0 top layer E B middle layer = Bottom layer

* not measured

0 5 2 7 37 LL 55 sample day

Figure 3. Course of various direct parameters (methanogenic capacity, amount of biomass on sand and volume of layers within the sludge bed) during start-up of R1 (A, C, and E) and R5 (B and D).

In Table I1 a comparison is made between the times at which the individual parameters showed a steep and persis- tent increase. From this, the onset of accelerated biofilm

formation can be timed for the various reactors. Taking into account the erroneous timing with the indirect parameters for R5 and R6, the biofilm production phase


Table 11. a steep increase was observed in the parameters.

Comparison of the times (days) after reactor star-up at which

Reactor Indirect parameters* Direct parametersb

R1 R2 R3 R4 R5 R6

20 37-45 35-42 35-42

0 0

<27 39 39 42 43 43

* Methane production rate, VFA-conversion rate and volume of the total

Methanogenic capacity, biomass on sand and volume of individual sludge bed.

layers.

started after about 40 days of operation in all instances, except for R1. In the case of R1, the onset of this phase was timed at day 20-27.

The sigmoid pattern observed in the course of biofilm formation in this study has also been observed recently for a 20-liter methanogenic FB-reactor4 and has been described for biofilm development under aerobic condition^'^ as well. During the period in which the various parameters increased only slowly, called the lag phase,4 the initial bacterial attachment to the surface of the support material is thought to take place. l9 This incipient colonization is followed by a period in which biofilm formation proceeds rapidly as a result of proliferation of the attached microor- ganisms. The second period is referred to as the biofilm production phase. In both periods, biofilm detachment may occur, mainly as a result of gas and liquid shear forces. These forces also determine the maximal biofilm thickness' and thus the plateau of the sigmoid curve, called the steady-state phase. In the experiments described here, granules in the different phases of biofilm formation were

found in separate layers in the sludge bed, probably as a result of their differing settling velocities. The plateau observed in biofilm formation in the present study may not only have been the result of an equilibrium between bacterial proliferation and mechanical shearing but also of the limited substrate supply towards the end of the experiments.

A number of sludge characteristics and the times of onset of biofilm formation determined in this study are compared in Table I11 to data reported for FB-reactor start-up at a larger scale under comparable operating conditions. In general, the biomass contents of sludge granules determined at steady-state were in the same range. The methanogenic activities measured in this study, however, were relatively high. This might be due to differences in the sludge activity tests employed. Biofilm formation accelerated after about 4 to 6 weeks of operation in all cases, except for R5 and R6 for which an instantaneous onset was indicated. The onsets were timed, however, using indirect parameters, which in the case of R5 and R6 gave erroneous results as discussed above.

Biofilm Composition

FB-granules were sampled from the top sludge layers when the various reactors were at steady-state in order to determine the bacterial composition of the newly developed biomass by means of scanning electron microscopy (SEM), activity measurement, and cofactor assay.

Examination of the samples by SEM, revealed that the biofilms in all instances mainly consisted of bacteria which morphologically resembled Methanothrix sp. ,20*2' whereas sludge from some reactors (indicated in Table V) appeared to contain Methanosarcina spp. 22 additionally. Apart from these acetotrophic methanogens, various other types of bacteria were observed, but none of these could be identified

Table 111. Comparison of start-up of methanogenic FB-reactors at different scales.

Operating conditions"

Effective Waste water Superficial Loading rate Sludge characteristicsb Onset of

the biofilm reactor VFA-COD liquid Hydraulic per amount Biomass on Methanogenic production phase' Calculated volume content velocity retention of sand sand activityd (days after using data

(L) (g COD/L) (m/h) time (h) (g COD/kg/d) (g VSS/kg) (g COD/g VSS/d) start-up) in/on

This study 0.46 0.4-2.0 9 1.7 37- 147 46/220' 5.1' 20; 37-45 R1 + R2 0.26 0.4-1.8 11 1.3 23-92 90/620' 4.3' 35-42 R3 + R4

0.26; 0.20 0.5-1.6 8-12 1.7; 2.2 53-212 82/225' 4.6' 0 R5 + R6 Lab scale 20 2 . 0 - 3 3 15-17 1.5 116187 370 3.0 25-30 ref. 4 Pilot scale 270 2.5-3.08 10-14 1.2-2.7 45- 120 110 1.8 27-44 refs. 1, 2, 18 Full scale 215000 2.28 15-20 1.6-3.4 82 120 2.0 100-12Oh ref. 3

a Determined over the effective part of the reactor. Measured at steady-state. As indicated by biogas production rate or VFA conversion rate. Measured in sludge activity tests with mixtures of acetate, propionate and butyrate as carbon sources. Average values for samples from middle and top sludge layer, respectively, for two independently operated reactors.

Pre-acidified yeast waste water. Reactor temperature was at 20-30°C during the first 70 days of operation, and at 37°C thereafter; all other reactors were operated at 35-37°C from

' Average activity of samples from the top layer for two independently operated reactors.

the start.


by morphology alone. As judged by epifluorescence microscopic ~bservation,’~ strongly ff uorescent Methanobacterium - type organisms” were present in all sludges.

The potential methanogenic activities of the various sludges measured on four different substrates are summa- rized in Table IV. The values of the methanogenic activities on each substrate varied somewhat between the different sludges. The methanogenic activities on acetate of all sludges, except of sludge from R6, were in the range of values reported for pure cultures of Methanothrix soehngenii, 1670 pmol CH,/g VSS/h,” but were lower as compared to the activity of Methanosarcina barkeri, 4130 pmol CH, g/VSS/h.” This finding is consistent with the SEM- observations that a Methanothrix sp. was the most abundant acetotrophic methanogen. The activities recorded with H,/CO, as the substrate were extremely low compared to the activity of Methanobacterium formicicum, 26000 pmol CH, g/VSS/h.25 This would indicate that only few hydrogenotrophic methanogens were present in the biomass, in contrast to the results of microscopic observation. How- ever, these values have to be interpreted with care, since it has been noted that insufficient transfer of hydrogen into the liquid phase in batch activity tests might lead to an un- derestimation of methanogenic activity on H2/C02. l 3 The

activities measured on propionate were comparable to values reported for UASB-sludges cultivated on mixtures of acetate and propionate or propionate alone, 120-220 and 340 pmol CH,/g VSS/h, re~pectively.~~,~’ With regard to the methanogenic activity on butyrate, no reference data were available in the literature.

Although the absolute values of the methanogenic activities on each substrate were at variance, in general, the various sludges were found to have a comparable relative substrate spectrum (Table IV). On average, the ratio of methanogenic activities was acetate : propionate : butyrate : H,/CO, = 60:5:30:5. From this it follows that the relative proportions of methanogenic and acetogenic bacteria were quite similar in all cases.

From the concentrations of specific methanogenic cofactors, measured in triplicate analyses (data not shown), the relative amounts (% of total biomass) of hydrogenotrophic and acetotrophic methanogenic bacteria in the newly developed biomass were calculated (Table V). The relative proportions of the acetogenic populations were derived from the total amounts of methanogenic biomass.

The calculated proportions show that the biofilm composition of all sludges was comparable, with Methanothrix sp. as the predominant organism (on average 72% of total

Tabie IV. carbon sources at steady s t aka

Potential methanogenic activities of newly developed fluidized bed sludges on various

Potential methanogenic activity (pmol CH,/g VSS/h) on the indicated substrate

Reactor/Inoculum Acetate Propionate Butyrate H,/C02 b Rl/effluent FB-reactor 1280 (56)” 480 (21) 525 (23) -

R3/sewage sludge (37°C) 2180 (59) 235 ( 6) 1200 (33) 85 (2) R4/sewage sludge (4°C) 2270 (74) 90 ( 3) 710 (23) 20 (1) RS/mature FE-granules 1400 (60) 55 ( 2) 720 (31) 170 (7) R6/combination of 1 and 2 780 (45) 90 ( 5 )

R2/effluent UASB-reactor 2185 (57) 200( 5 ) 1120 (29) 325 (8)

860 (50) -

a Fraction (percentage) of the sum of the activities on each of the four substrates, the ratio of the fractions is the relative substrate spectrum.

Activity test not performed.

Table V. sludges calculated from cofactor contents.

Relative amounts of methanogenic species and nonmethanogens in the various FB-

Relative proportions (% of total biomass)

Reactor/Inoculum Methanobacterium Methanothrix Methanosarcina Others“

Rl/effluent FB-reactor 6.2 78.6 0.7 14.5 (-)‘ R2/effluent UASB-reactor 14.5 60.6 9.1 15.8 (+++) R3/sewage sludge (37°C) 12.1 79.2 2.3 6.4 (+)

RS/FB-sludge 6.4 71.9 12.1 9.6 (+++) R4/sewage sludge (4°C) 9.6 78.9 1.8 9.7 (+)

R6/FB and sewage sludge 9.5 59.5 1.2 29.8 (-)

a Calculated by subtraction of the sum of the relative proportions of methanogenic bacteria from 100% biomass.

Average values based on cofactor contents (data not shown). Abundance of Mefhanosarcina spp. observed in SEM-preparations: - = not detected; + and

+ + + are 1-5% and 10-15%, respectively, of the estimated total methanogenic population at the surface of the granule.


biomass). Using Methanobacterium formicicum as reference, substantial amounts of hydrogenotrophic methanogens were estimated (average 10%). Methanosarcina spp. were found to be present in all instances, but were most abundant in sludge from reactors 2 and 5. This is in good agreement with the results of microscopic observation. Both types of acetotrophic methanogens were also observed during start-up of pilot plant' and full scale* methanogenic FB- reactors on preacidified yeast waste water. From the relative proportions of the various methanogens it follows that only a small part of the biomass consisted of nonmethanogens. These other bacteria, amongst others the propionate and butyrate consuming acetogenic bacteria, comprised on average 14% of the total biomass (Table V).

In summary, all measurements performed indicated that the relative amounts of acetogenic and methanogenic organisms in the newly developed biomass were very similar for the various sludges. Thus, it appeared that the different types of inoculum used in this study did not influence the biofilm composition significantly.

CONCLUSIONS

Under the experimental conditions of the laboratory set- up, biofilm development was successfully obtained with various different types of inoculum. The characteristics of the FB-granules which had developed were comparable with granules obtained in pilot plant and full scale FB-reactors. Thus, the set-up employed in this study appears to be suited for a systematic investigation on factors affecting biofilm formation in methanogenic FB-reactors.

In the present study, FB-reactor start-up proceeded in three consecutive phases, viz lag phase, biofilm production phase and steady-state phase, irrespective of the type of inoculum applied. This pattern reflected the overall rate of colonization and biofilm production for granules present in different layers within the sludge bed. The pattern was assessed most validly by parameters on sludge level, whereas the more indirect parameters gave a false reflection of the course of biofilm formation on the sand particles.

The biomass composition of the newly developed granules analyzed in the stationary phase was found to be similar w i t h all types of inoculum. A predominance of Methanothrix-like acetotrophs was noticed in all newly developed sludges.

The type of inoculum did not significantly affect either the course of biofilm formation or the biofilm composition during start-up with one type of waste water. This indi- cates that the similarity in environmental conditions, par- ticularly nutrients, in the reactors is of greater importance in determining the microbial composition of the system than the variation in the inoculum. An influence of the types of volatile fatty acids in a waste water on biofilm composition has recently been found employing the experi-

mental set-up described here. A detailed report will be given elsewhere.

This investigation was supported in part through a financial grant by Gist Brocades BV Delft, The Netherlands.

References

1. J. J. Heijnen, in Proc. Eur. Symp. Anaerobic Waste Water Treat- ment, (Noordwijkerhout, The Netherlands, 1983) p. 259.

2. J. J. Heijnen, A. Mulder, W. Enger, and F. Hoeks, in Proc. Water Treatment Conference AQUATECH '86 (Amsterdam, 1986) p. 161.

3. W. A. Enger, W. M . A . van Gils, J . J . Heijnen, and W. A. A. Koevoets, in Proc. Water Treatment Conference AQUATECH '86 (Amsterdam, 1986) p. 297.

4. A. Mulder, in Proc. Water Treatment Conference AQUATECH '86 (Amsterdam, 1986) p. 669.

5 . J. Dolfing, Waf. Sci. Tech. 18, 15 (1986). 6 . P. Huysman, P. van Meenen, P. van Assche, and W. Verstraete in

Proc. Eur. Symp. Anaerobic Waste Water Treatment, (Noordwijker- hout, The Netherlands, 1983) p. 107.

7. M. S . Switzenbaum, K. C. Scheuer, and K. E. Kallmeyer, Biotech- nol. Lett. 7 , 585 (1985).

8 . W.E. Balch, G.E. Fox, L.J. Magrum, C.R. Woese, and R . S . Wolfe, Microbiol. Rev. 43, 260 (1979).

9. T. J. Hutten, M. H. de Jong, B. P. Peeters, C. van der Drift, and G. D. Vogels, J . Bacteriol. 145, 27 (1981).

10. H. J. Gijzen, K. B. Zwart, P. T. van Gelder and G. D. Vogels, Appl. Microbiol. Biotechnol. 25, 155 (1986).

11. Anonymous, Standard Methods for the Examination of Water and Waste Water (American Public Health Association, New York, 1975).

12. L. G. M. Goms, I. M. A. van Deursen, C. van der Drift, and G. D. Vogels, Appl. Microbiol. Biotechnol., 29, 95 (1988).

13. J . Dolfing and W. G. B. M. Bloemen, J . Microbiol. Methods. 4, 1 (1985).

14. D. Valcke and W. Verstraete, in Anaerobic Digestion 1981, (Elsevier Biomedical Press, Amsterdam, 1982) p. 385.

15. L.G. M. Goms and C . van der Drift, in Biology of Anaerobic Bacte- ria, (Elsevier Science Publishers, Amsterdam, 1986) p. 144.

16. L. G. M. Goms, C. van der Drift, and G. D. Vogels, J . Microbiol. Methods 8, 175 (1988).

17. L. G. M. Goms, H. A. Kemp, and D. B. Archer, Biomass, 14, 195 (1987).

18. J. J. Heijnen, in Proc. Eur. Symp. Anaerobic Waste Water Treat- ment, (Noordwijkerhout, The Netherlands, 1983) p. 176.

19. J . D. Bryers and W. G. Characklis, Biotechnol. Bioeng. 26, 2451 (1982).

20. B. A. Huser, K. Wuhrmann, and A. J. B. Zehnder, Arch. Microbiol. 132, l(1982).

21. G. B. Patel, Can. J . Microbiol. 30, 1383 (1984). 22. R. A. Mah and M. R. Smith, in The Prokuryotes, Vol 1. (Spnnger-

23. H. J. Doddema and G . D. Vogels, AppZ. Environ. Microbiol. 36, 752

24. J. A. Krzycki, R. H. Wolkin and J. G . Zeikus, J . Bacteriol. 149, 247

25. N. L. Schauer and J. G. Ferry, J . Bacteriol. 142, 800 (1980). 26. J. Dolfing and J. W. Mulder, Appl. Environ. Microbiol. 49, 1142

27. E. ten Brummeler, L. W. Hulshoff-Pol, J. Dolfing, G. Lettinga, and

Verlag, Berlin, 1985) p. 948.

(1978).

(1982).

( 1985).

A. J. B. Zehnder, Appl. Environ. Microbiol. 49, 1472 (1985).


biofilm development in laboratory methanogenic fluidized bed reactors

Documents