~ Pergamon War. Sci. Tech. Vol. 33. No 7. pp. 157-164. 1996.Copyright © 1996 IAWQ. Published by Elsev,er SCIence LId

Pnnted on Great Bnlam All nghts reserved.0273-1223/96 S15'00 + 0'00

PH: S0273-1223(96)00351-4

MELBOURNE WATER'S WASTEWATERTREATMENT LAGOONS: DESIGNMODIFICATIONS TO REDUCE ODOURSAND ENHANCE NUTRIENT REMOVAL

Brian Hodgson* and Peter Paspaliaris**

'" Department ofMicrobiology, University ofMelbourne, Parkville, 3052. Australia"'''' Melbourne Water, /4/625 Little Collins Street. Melbourne 3000, Australia

ABSTRACf

Some properties of 3 "new style" wastewater treatment lagoons. 115E, S5E and 25W at the MelbourneWater, Western Treatment Plant (WTP) treating some 250 megalitres (ML) of untreated wastewater each dayare described. There is a potential residence time for each of 120 days and each consists of a sequence of upto II ponds. Pond I has an anaerobic reactor of 90, 150, and 150 ML respectively and Warmens floatingaerators are installed on ponds I and 2 of lISE and 2SW and pond I of SSE. BODS values of less than 50 areachieved by the end of pond 2 and these together with the installation of the HOPE cover on 115E haveeffectively reduced odour emissions. Nitrogen is removed by ammonification followed by eithernitrification/denitrification, or algal growth which is grazed by zooplankton. Since the introduction of theaerators. chemolithotrophic ammonia oxidising bacteria (CAOB) are more frequently exposed to theinhibitory action of UV light, and therefore nitrification is more sporadic. Turbidity of the water may playasignificant role in protecting the CAOB from UV light. The lagoons have the potential to produce an effluentWith inorganic-N levels of less than 2 mgIL, a BODs of less than 50 mgIL and low levels of algae. Thecovered anaerobic reactor can in each case produce up to 20,000 cubic metres of gas each day comprising of80% methane. Methane will be used to generate electricity, and the zooplankton generated by feedmg onalgae will be harvested to provide food for fish fry. Copyright © IAWQ 1996. Published by Elsevier ScienceLtd.

KEYWORDS

Aerators; algae; anaerobic reactors; HOPE covers; lagoons; methane; nitrification; odour; UV light;zooplankton.

INTRODUCTION

Melbourne Water's Western Treatment Plant (WTP) for wastewater, at what was previously called theWerribee Farm, is on 10,500 hectares of land and uses 3 treatment processes, grass filtration in winter, landfiltration in summer and lagooning year round. The land and grass filtration processes support livestockfarming with 15,000 cattle, 60,000 sheep and 5,000 goats, generating an annual income of $3 million. TheWTP services 1.4 million people and receives 80% of Melbourne's industrial wastewater with an averagedaily flow of 500 megalitres (ML) that can reach up to 1,600 MUday in wet weather. The treated waterflows into Port Phillip Bay.

157

158 B. HODGSON and P. PASPALIARIS

The lagooning process was introduced in 1936 to act as a polishing system for the grass filtration process inpeak. wet flow conditions. This was a 271 hectare pond that was later renamed Lake Borrie and enlarged to425 hectares. This is now part of a wildlife reserve of international significance for water birds. Since thenthe lagooning process has been constantly developed with new lagoons now treating up to 65% of the totalannual flow into the WTP.

In June 1986 the first of the "new style" lagoons. lISE. was commissioned in an attempt to reduce nutrientlevels in the effluent. in particular N-content, as biological activity in the Bay is N-limited. This one lagoonhas II ponds each about I kilometre long and a total operating volume of 4,23S ML about 4 times thevolume of the previous largest lagoon. Some of the properties of this and more recently commissionedlagoons are described in the next sections.

MATERIALS AND METHODS

Sampling. Water samples were taken according to standard methods (Anon 1976).

Chemical analysis. Assays for BOD5• suspended solids (SS). total Kjeldahl nitrogen (TKN), total dissolvedsolids (TDS). pH, chlorophyll-a, sulphide. total organic carbon (TOC). dissolved oxygen (DO), werestandard methods (APHA 1985). Ammonia nitrogen (NHrN). nitrite nitrogen (NOrN), nitrate nitrogen(NOrN), ortho-phosphorus (Po). and total phosphorus (PT) was by flow injection analysis with a Tecatormachine. model No. SOZO, Tecator AB. Hoganas, Box 705-263-01 Sweden. The methods are described inthe "Tecator" application notes.

Enumeration or bacteria. E. coli and coliform counts were performed by the multiple (5) tube methodaccording to standard methods (Anon 1982) with reference to probability tables (APHA. 1971). Salmonellaspp were enumerated by the multiple (3) tube method (Anon, 1982) and standard probability tables (APHA,1971). Samples were first diluted in Ringer's solution to 10-1• 10-2• 10-3 and 10-4• 10 mL of undilutedwastewater was added to each of 3 tubes of 10 mL of double strength buffered peptone water (BPW), I mLto each of 3 tubes of 6 mL of single strength BPW and I mL of each dilutIOn added to each of 3 tubes of 6mL of BPW. These were incubated for 18 h at 3S"C as a pre-enrichment step. For enrichment, 0.3 mL fromeach BPW culture was added to 6 mL of Rappaport Vassiliadis (RV), (Harvey and Price. 1983) broth (OxoidLtd.. London, England) and incubated for up to 72h at 4Z"C. The RV medium is similar to that described byVassiliadis et al. (1976) except the peptone used is soya peptone which enhances the growth of Salmonellaspp (van Schothorst and Renaud, 1983). For isolation a loopful of each enrichment culture was subculturedon to xylose/lysine/desoxycholate agar (XLD). Oxoid LId.• London, England) and Salmonella/Shigella agar(5S. Oxoid Ltd.• London. England) and incubated for 48h at 37"C. At least one isolated colony was thenpicked from each plate. tested for purity on a nutrient agar plate and confirmed as a Salmonella spp withurea broth, triple sugar iron agar (TSI. Oxoid Ltd.• London. England) and lysine decarboxylase broth (Taylormodification. Oxoid Ltd.• London, England). Initial serological confirmation was by a slide agglutinationtest using polyvalent "0" antisera (Wellcome Diagnostics, Dartford. England). Complete serologicalidentification was performed on selected isolates by the Microbiological Diagnostic Unit, The University ofMelbourne. according to Edwards and Ewing (1972). Faecal streptococci were enumerated by the standardmultiple tube (S) method for enterococci (Method 910 A. Multiple-tube technique APHA, 1985).Chemolithotrophic ammonia oxidising bacteria (CAOB), were enumerated by a standard multiple tube(3) procedure (APHA, 1971) with one of 3 different media. Medium-I was that of Skinner and Walker(1961) supplemented with a trace metal mix. Medium-2 was similar to medium-I except that the ammoniumsulphate level was reduced from O.S to 0.2 gIL. Medium-3 was filter sterilised lagoon water, normally fromIhe 45W lagoon that contained about 30 mg NH3-NIL and demonstrated good year round nitrification.Filtration through coarse fibre glass filters was followed by filtration through 0.45 mm and 0.22 mm Gelmanmembrane filters. 0.5 mL samples serially diluted in Ringers solution to 10-8 were each added to 4.5 mL ofmedium. incubated for up to 8 weeks at 2S"C and tested for the presence of nitrite by a spot test (Anon.1995). Cultures showing no nitrite were treated with Zn-dust and retested for nitrite. Any culture giving apink colour in either test was regarded as positive and counts were determined by reference to the standardprobability tables (APHA. 1971).

Melbourne water's wastewater treatment lagoons 159

Bacteriophage. Coliphage and Pseudomonas aeruginosa phage were enumerated using standard methods(Anon, 1986). Host strains were E. coli MU-209, and Pseudomonas aeruginosa MU-353 both from theculture collection at The University of Melbourne, Department of Microbiology.

Odour emission rate measurements. A device known as the "floating odourhood" (FOH) was developedby Melbourne Water to measure odour emission rates (base area 0.071 m2 and volume 20 L). Odour free airsupplied from a cylinder of compressed medical grade air is bled into one end of the FOH and air samples tobe measured from the other end. A pump draws a vacuum on a drum containing a Milanter sampling bag.This draws air from the FOH into the bag. By having positive pressure from the compressed air bottle, and asuction from the pumps, the pressure inside the FOH can be maintained at ambient, so the emission rate isnot affected by pressure differences. The odour strength of the air samples is assessed by an odour panelconsisting of 5 members. Each is supplied with gas from the bag by means of a divergence olfactometerdesigned to measure how much an air sample needs to be diluted so that only 50% of the population can nolonger detect an odour. Air from the sample bag is mixed with medical grade odour free air maintained at aflow rate of 125 Urn. An initial odorous air flow is chosen so that all panellists can smell an odour. Theodorous air flow is decreased with steps of 50% until less than half the panel can detect an odour.Presentation of odorous air is randomly interspersed with blanks (odour free air) to guard against panellistsguessing. ODU values = 125 + (Sample flow, Urn) / (Sample flow, Urn). The number of dilutions is plottedagainst % response from the panel on a logllinear graph. The number of dilutions for the 50% response isthen read off.

RESULTS AND DISCUSSION

115E la&oon properties

All ponds of the lagoon have been sampled at their outflow every 2 weeks. The average daily flow rate ofuntreated wastewater into the lagoon is 60 ML. During a 1 year period beginning June 88, for pond 10, theaverage NH3-N level was 14, NOTN 9.5, BODs 61.5 mglL. During winter (i.e. June-Sept) these values were21,8, 105; and in summer (ie Dec-March) 0.8, 14,27 respectively (Table I). Nitrogen was being removedfrom the system by ammonification followed by nitrification/denitrification. This occurred mainly insummer when the water temperature was at least IO'C higher than in winter. BODs removal was also moreeffective in summer but as can be seen from the standard deviations. occasional high values, whichinvariably coincided with an algal bloom, upset the averages. Algal blooms occurred sporadicallythroughout the year but mainly in summer when they also coincided with higher NOrN levels.

Table 1. Average water quality values for 115E pond 10

mgIL

Time period NH3-N N03-N BOD5

Before aeration

1/6/88 to 31/5/89 14± 12 9.5±5.5 61.5 ± 61

1/6/88 to 28/9/88 21 ± 8.5 8±5.5 105±707/12/88 to 22/2/89 0.8 ± 1.0 14±3 27±16After aeration

6/6/90 to 29/5191 16±7 1.5 ± 1.5 8.S±4.56/6/90 to 26/9190 20±3.S I.S ± I.S 8±4S/I2190 to 27/2/91 13.5 ±7 0.5 ±0.3 10± 5.S

160

Effect of aerators on water Quality

B. HODGSON and P. PASPALIARIS

As a result of the rapid urbanisation of the land surrounding the farm during the late 1980s, there wasincreasing pressure from residents for measures to reduce odour levels. Odours were generated in theanaerobic ponds Le. I and 2 of the lagoons, the first 50m of the grass filtration bays, the sedimentation tanksand sludge digestion ponds. Odourhood measurements at the sides of the anaerobic pond gave odour dilutionunit (ODU) values of about 2,000. In 1989 Warrnens floating aerators (30 kW + 15 kW) were installed in thelast 200m of pond I and along the length of pond 2 of 115E lagoon. This had a significant effect on thequality of the effluent. Over a I year period beginning June 90, of pond 10, the average NH3-N level was 16,NOrN 1.5, BODS 8.5 mgIL. In the same pond in winter these values were 20, 1.5, and 8, and in summer,13.5,0.5, and 10 respectively (Table I). Although the BOD values had reduced significantly and varied littleduring the year this correlated with no significant algal blooms and very little nitrification. The BODS ofpond I outflow reduced from about 150 to 50, and DO levels of 2 mgIL were frequently recorded. This wascorrelated with a reduction in odourhood measurements at the side of pond I to ODU values in the region of160.

Factors affectin~ nitrjfication

It was important to understand the reasons for the reduction in nitrification. In our experience goodnitrification requires populations of CAOB of greater than 108/100 mL as measured by the 3 tube MPNmethod using the medium of Skinner and Walker (1961), with an initial reduced NHrN level of 42 mglL,after an 8 week incubation. Levels of CAOB in the water column of any of the 115E ponds following theinstallation of aerators rarely exceeded 104/100 mL, even though levels of CAOB greater than 108/g dryweight were frequently found in the sediment. Water samples from any of the ponds I to 4 incubated inflasks in a shaking incubator at 20'C produced rates of NHrN oxidation in excess of 5 mglLlday which wassufficient to remove all NHrN within 10 days. At this time CAOB counts were greater than 10 10/100 mL(Smith, 1992). The low level of CAOB in the pond water column was shown not to be due to the presence ofinhibitory chemicals in the wastewater. CAOB were therefore present in ponds I to 4 and gIven theappropriate conditions should grow and remove most NH3-N in anyone pond as the residence time for eachis 7,9,8 and 9 days respectively.

Experiments with pond water in quartz and pyrex flasks incubated at the surface of pond 4 showed that UVlight was an inhibitor of CAOB, and as the water in the ponds was totally mixed by wind action during theday most bacteria in the water column would be exposed to UV light at some time. Although recoveryoccurs in the dark, this is obviously not sufficient time to allow rapid metabolism and growth. Recentexperiments have established that good nitrification rates can be achieved in quartz flasks protected fromUV light by shade cloth allowing either 30% or 50% of incident UV to penetrate (Dommisse, 1994a). Weassume that prior to the introduction of the aerators water quality occasionally protected the CAOB from theUV light. At one time because of the consistent association of algal blooms with periods of goodnitrification we thought that the algae could protect the CAOB from UV light. Experiments in quartz flaskswith algae and nitrifying populations have so far shown no such effects. Algae and CAOB compete for NH3­N with some algae utilising NOrN, the product of nitrification. Algae utilise the inorganic nitrogen forgrowth and therefore produce more organic nitrogen, while nitrifying bacteria produce mostly NOrN andsome cell material. Algae are grazed by zooplankton. From experiments measuring total heterotrophiccounts and CAOB counts there is no evidence for the preferential grazing of CAOB. We conclude thatconditions that favour the growth of CAOB also favour algal growth.

Die-off of potential patho~ens

One important outcome of wastewater treatment is that potential pathogens are removed before treated wateris released into the receiving water. In this case, Port Phillip Bay is salt water, so most human pathogenswould find it difficult to survive. However shellfish and fish could act as carriers and a reservoir of infectionfor humans. Harvesting of shellfish and fish is prohibited in areas adjacent to the WTP but this is difficult toenforce and fish are very mobile. Experiments on the survival of potential pathogens and indicatororganisms in the lagoons have shown good die-off (Table 2). In almost all cases, background levels of these

Melbourne water's wastewater treatment lagoons 161

bacteria were achieved by pond 5. The E. coli and Enterococcus spp. observed in later ponds may be due tothe presence of birds in these ponds.

Table 2. Die-off of potential pathogens from liSE lagoon

17112190 orgs./lOOml

Salmonella Enterococcus sp.

sp.

Input

Pond 1

Pond 2

Pond 3

Pond 4

Pond 5

Pond 6

Pond 7

PondS

Pond 9

Pond 10

E. coli

1.3 x 107

2.5 x 1057 x 104

250

250

25

35

50

25

2017

43002304

4

<2<2<2<2<2<2<2

7.4 x 106

7.6 x 1051.3 x 104

850

740473250

32061

6

At another time, during winter, the die-off of viruses was modelled with coliphage (Table 3). Coliphage wasnot detected after pond 5 even though host E. coli was still present, so it would be reasonable to assume thatviruses such as HepA, for which no host is available, would be effectively removed. This was one of the fewexamples when E. coli levels did not reach background levels until pond 8. Following the introduction ofaerators the results with the CAOB indicate that exposure of water microbes to UV light and thereforekilling has been facilitated. Die-off of indicator organisms was then more rapid.

Table 3. Die-off of E. coli and coliphage

Input

Pond 1

Pond 3

Pond 5

Pond 7

Pond 9

Pond 11

Coverjnf; of the anaerobic reactor

E. coli 1100 rnl

1.3 x 107

S xl06

2.5 x 106

8008050

50

Coliphage 1rnl

380250

54

18<1

<1<1

The second major modification to the liSE lagoon was made in 1992 with the installation of a floating 2.5mm thick HDPE cover over the first 100 m of pond I. This is now known as the Anaerobic Reactor with anarea of 3 Ha, a depth of 3 m and a volume of 90 ML (Table 4). Gas accumulates under this cover at a rale of5,000 to 20,000 cubic metres each day and comprises 80% methane. This methane will be used to generatepower at the WTP. We have not observed that this installation has had an effect on the rest of the lagoonprocess. At present, wastewater enters the anaerobic reactor with a BODs of 450, leaves it at 170, and afterthe aerators leaves pond I at 50 mglL.

162 B. HODGSON and P. PASPALIARIS

ADA POND 1

LAGOON

100m

DIAGRA....ATIC LAYOUT

.. . : ~: . ... . . ,~

¢:J "',"~

POND 3 I~

"I"¢:raND 5

vv

el)

PONO z-+ :I-=--=---=:"--=--=---=:"--=-~,-j

POND' POND POND' POND • POND 10

POND S~ t SS' lACDD'¢:JPONO.

DIAGRAIolIolATIC LAYOUT

POlIO 7~!!!Illm

¢:lI'ONO •

"'NO .~

¢:>I'OMO '0

115' ....GOO, tDIAGRAMt.4ATIC LAYOUT

SOQm

1'0100 11 ¢:J

"'NO 10 ¢:J

FIgure I. Layout of new lagoons.

Melbourne water's wastewater treatment lagoons 163

Table 4. New lagoon dimensions

Total Anaer. Total Anaer. Aerated Anaer. Depth 01lagoon Reactor operating reactor pond reactor rest of

area area vol. vol. vol. depth lagoonHa Ha ML ML ML In m

llSE 200 3.0 4235 90 360 3 2.1SSE 287 3.3 6288 ISO 665 6 1.92SW 271 3.3 6189 ISO 710 8 2.0

SSE and 25W la~oons

Lagoon 55E was commissioned for full scale operation with aerators in May 1992. The only significantdesign difference from lISE was the depth of the anaerobic reactor which was increased to 6 m (Fig. 1).This produced an anaerobic reactor with a volume of 150 ML. The operating volume of this lagoon is 6,288ML and untreated wastewater enters at a rate of 100 MUday. The anaerobic reactor will be covered afterresults from experiments with the lISE anaerobic reactor are completed and analysed. During a 1 yearperiod beginning July 93, for pond 10, the average NH3_N level was II, NOrN 3 and BODs 10 mglL. Inthe winter of this period, from 9/6/93 to 22/9/93 the average values were 30, 3, and 19 respectively andvalues of NH3-N of 37 were recorded. In contrast in the summer of this period, from 15/12/93 to 15/2/94 theaverage values were 1.3, 0.5, and 8 (Table 5). Following the period of negligible nitrification during winter,an NHrN level close to I was first observed on 3/11/93 with an NOrN level of 7 mgIL, i.e. nitrificationhad resumed. On 24/11/93 an NHrN level of 1.7 was recorded in ponds 4 and 5 with N03-N levels of lessthan 4. At this time chlorophyll-a values greater than 1,000 were present in ponds 4, 5, 7 and 9 whichindicated large algal blooms. These blooms reduced considerably within a month and up to April 94 levelsof NH3-N, NOrN and algae remained low. We believe that the inorganic-N is being converted to algaewhich is being grazed by zooplankton. Some ammonification occurs as a result of this process and supportslimited nitrification. As a result in pond 4 there can be an NHrN level of 1.7 an N03-N of 3.6 andchlorophyll-a of 1,020 while pond 8 has values of 4.8,8.4 and 405 respectively.

Table 5. Average water quality values for 55E pond 10

Time period

07/07193 - 0110619409/06/93- 22/09193

IS/12/93 - 15/02194

11.2 ± 12.6

30.3±S.7

1.3 ± 1.4

mg/l

N03-N

2.6 ± 3.1

2.6±2.9

O.S ±0.6

BODS

10±9

19± 13

8±3

Lagoon 2SW was commissioned for full scale operation with aerators in Nov 1994. This is now receivingabout 90 MUday has an operating volume of 6,189 ML and an anaerobic reactor volume of ISO ML with adepth of 8 m (Fig. I). The design incorporates a facility for the 2,236 ML pond 3 to be by-passed (Fig. I).This was to be used as a "nitrification loop" where retention times could be extended for slow growingnitrifiers to multiply and be used to inoculate pond S. The design was completed prior to the findings of therole for UV light. Wastewater first flowed into this lagoon in March 1993 and samples were first taken frompond 10 in August 1993. Since then ponds 3 and 4 have consistently had large algal blooms withchlorophyll-a levels as high as 1,750 in April 94. A < 0.1 mgIL NH3.N level was first recorded in pond lOinOctober 93 with a NOrN level of 13.7 mgIL.

By January 94 an NHrN level of 0.1 was recorded in pond 6 with a NOrN level of 18.3 mgIL. Levels ofNH3-N of less than I were recorded in pond 6 until 4/5/94 but during the extremely high algal bloom inApril the NOrN levels were less than 0.1 mgIL. At this time algae were using the NOrN. However at othertimes e.g. 8/2/94 pond 5 had an NHrN level of 0.1 a NOrN level of 11.8 and a chlorophyll-a level of1.410. At different times different algae predominate and invariably different algae predominate in ponds 3and 4 compared to those in ponds 7 and 8 (Cartwright, 1994). Some algae use the N03-N produced by the

164 B. HODGSON and P. PASPALIARIS

nitrifying bacteria more readily than others (Dommisse. 1994b). Nitrification had ceased by 27n/94 and wasfirst observed again in pond 10 on 9/11194 at which time the flow rate was increased to the design rate of 90MUday.

CONCLUDING REMARKS

Although some progress has been made in understanding the processes that operate in the "new lagoons"there is still a lot to be learned before adequate control models can be developed. It is obvious that thelagoons can operate effectively throughout the year i.e. produce an effluent with inorganic-N levels of lessthan 2 mg/L with a BOD less than 20 mg/L and low levels of algae. More information is required on thenitrifying bacteria. There is evidence that nitrifying populations are heterogenous and change withenvironmental conditions. Is there one type best suited to our conditions? What is the best design for theanaerobic reactors to optimise conversion of organic-C to methane? Is there a need for a "nitrification loop"such as exists in 25W? Can we optimise the production of algae to provide feed for zooplankton which canbe harvested for fish-fry? These questions can only be answered by further work.

REFERENCES

Anon (1976). Notes on water sampling. VictOrian EPA. Laboratory services branch. May 1976.Anon (1982). The Bacteriological Examination of Drinking Water Supplies. Reports on Public Health and Medical subjects

No.71. (Bulletin No. 71).Anon (1986). Laboratory manual for Agricultural Microbiology Course. Department of Microbiology, The University of

Melbourne.Anon (1995). Techniques Manual, Department of Microbiology. The University of Melbourne. p. 137. test-I.APHA (1971). Standard Methods for the Examination of Water and Wastewater. 13th. edn. American Public Health Association.

Washington. DC.APHA (1985). Standard Methods for the Examination of Water and Wastewater. 16th. edn. American Public Health Association.

Washington. DC.Cartwright, D. (1994). Aquatic biologist. Western Treatment Plant. Melbourne Water. Personal communication.Dommisse, M. (1994a). A preliminary investigation into the use of shadecloth to optimise nitrification In wastewater lagoons.

Report prepared for Melbourne Water 1994.Dommisse. M. (1994b). Aspects of ammoma removal in the treatment of wastewater by the lagoomng process at Melbourne

Water's Western Treatment Plant. Honours thesis (Microbiology) The Umversity of Melbourne.Edwards. P.R. and Ewing W.H. (l972)./dentlfication ofEnterobacteriaceae. 3rd edn. Burgess Publishing Company. Minneapolis.

Minnesota.Harvey. R.W.S. and Price, T.H. (1983). A comparison of two modifications of Rappaport's enrichment medium (R 25 and RV) for

the isolatIOn of salmonellas from sewage polluted natural waters. J. Hygiene. 91. 451-468.Skinner. F.A. and Walker, N. (1961). Growth of Nitrosomonas europaea In batch and continuous culture Archives fur

Mikrobilogie, 38. 339-349.Smith. A. (1992). An investigation of the factors affecting the nitrification processes occurring In the wastewater lagoons at the

Melbourne Water Wembee Treatment Complex. Honours thesis (MicrobIOlogy). Umverslty of Melbourne.van Schothorst. M. and Renaud. A.M. (1983). Dynamics of salmonella isolation with modified Rappapon's medium (RIO). J.

Applied Bacteriology, 54. 209-215.Vassiliadis. P., Pateraki, E., Papaiconomou N.• Papadakis, lA. and Trichopoulos. D. (1976). Nouveau procede d'ennchissement

de salmonella. Ann. Microbiologie (Paris). 1278. 195-200.