FR :

La désodorisation physico-chimique en stations d'épuration s'effectue généralement par lavage basique oxydant pour piéger les espèces soufrées réduites telles que H₂ S ou CH₃ SH. L'utilisation du peroxyde d'hydrogène n'est pas encore répandue en comparaison de celle du chlore. Cette étude a été menée afin de déterminer le comportement de H₂O₂ en fonction de la composition de l'eau de lavage. L'influence des paramètres : concentration en métaux (fer, manganèse, cuivre et zinc), pH, [H₂O₂], [CO₃^2-], [HS^-] a été étudiée en utilisant un plan d'expériences. La décomposition de H₂O₂ et la concentration de radicaux libres ont été mesurées pour chaque expérience. En présence de métaux, un pH élevé et une forte concentration en peroxyde sont les deux paramètres principalement responsables d'une forte décomposition. Cette décomposition serait accompagnée d'une production de radicaux avec [HO°]_max =10^-13 M. Cette valeur mesurée de radicaux dans le milieu n'explique qu'une petite part de la décomposition de peroxyde observée. Par conséquent, la majorité de la décomposition est due à des réactions soit à la surface des oxydes, soit en solution avec les cations dissous. Le mélange de métaux et de carbonates à pH 10,5 présente un effet de synergie sur la décomposition de H₂O₂. Ces résultats démontrent que malgré le pouvoir oxydant des radicaux HO° formés, l'utilisation de H₂O₂ en désodorisation ne sera possible qu'avec l'ajout de stabilisant.

EN :

Deodorization of wastewater treatment plants involves the elimination of molecules such as NH₃, amines and sulphur compounds like H₂ S and mercaptans. In classical physico-chemical processes, NH₃ and amines are trapped in acid solution by washing air in a scrubbing tower, while sulphides are eliminated in basic oxidising solutions. The oxidant usually used is sodium hypochlorite. Elimination of sulphides and organosulphides generally demands two scrubbers: one at pH 9 and the other at pH 11. Because chlorine in deodorization generates the formation of organochlorinated species, it should soon become necessary to replace this oxidant in order to avoid the formation of such compounds. The present study follows the behaviour in wash conditions not of chlorine, but hydrogen peroxide, in order to discover the deodorization capacity of this molecule.

The kinetics of H₂ S oxidation by H₂O₂ are well known; the constant is given by

log k = 12.04 - (2641/T) - 0.186 x pH (Millero et al., 1989).

Unfortunately H₂O₂ shows strong decomposition in alkaline medium, due to the presence of metals and carbonates in the solution. Initiating a homolytic reaction results in the decomposition of peroxide. However, increasing the concentration of free radicals may improve H₂ S oxidation and consequently, the efficiency of the process.

To better understand the behaviour of H₂O₂ in wash conditions, various parameters were studied, namely pH (9 and 10.5), [H₂O₂] (1 and 5 g L^-1), metal concentrations (iron, manganese, copper and zinc) (20 and 200 µg L^-1), [CO₃^2-] (0 and 100 mg L^-1) and [HS^-] (0 and 2 mg L^-1). Four experimental designs, one for each metal, were employed to reduce the number of experiments and benefit from statistical laws. H₂O₂ decomposition and HO° concentration were measured and empirical equations established. All experiments were performed in closed-batch reactors with ultra-pure reactants and water. Measurements of HO° concentrations necessitated the addition of atrazine to the solution. The oxidation of this pesticide by HO° is well known. Using atrazine concentrations measured through time, the HO° concentrations were calculated according to the equation

ln ([Atz]₀/[Atz]) = k[HO^∘]t

with k=2.1 × 10⁹ M^{-1 s-1} (De Laat et al., 1997). Oxidation of atrazine was halted by extraction onto a C_t18 Sep-Pack resin and samples were analysed by liquid chromatography.

The results showed that in the presence of metals H₂O₂ decomposition was maximal at high pH and with high peroxide concentrations. The decomposition was accompanied by HO° production. However, the presence of metals generated the decomposition of H₂O₂ with a reduced production of free radicals compared with ultra-pure water, which indicates that metal oxides were not only decomposition catalysts, but also radical inhibitors. Comparison of simplified radical decomposition, calculated according to the equation

([H₂O₂]/[H₂O₂]₀)=e^-k[HO∘]t,

and observed decomposition showed that under these conditions H₂O₂ consumption was mainly due to metal reactivity. Nonetheless, increasing iron and copper concentrations from 20 to 200 µg L^-1 did not modify the decomposition rate of H₂O₂. For this reason we postulate a Fenton-like reaction between H2O2 and dissolved metals in which concentrations are determined by solubility products. It follows that the kinetics of H2O2 decomposition can be summarised by r=-k₁ [oxide][H₂O₂] - k_2[ Mⁿ⁺][H₂O₂] - k₃ [HO°][H₂O₂], with [metal]_Tot =[Mⁿ⁺] + [oxide] and, in the case of Cu and Fe, k₁ [oxide][H₂O₂] << k_2[ Mⁿ⁺][H₂O₂].

To conclude, the addition of four metals with [CO₃^2-]=1 g L^-1 at pH 10.5 produces a synergetic effect, resulting in a much faster decomposition. These conditions, unfortunately, resemble deodorization conditions. The use of a stabiliser that inhibits not only free radicals but also decomposition catalysts is therefore necessary for deodorization.

FR :

Les piscicultures en circuits semi-fermés sont confrontées au problème de l'élimination permanente des matières en suspension (M.E.S.) et des substances azotées. Les procédés conventionnels utilisés pour retenir les M.E.S. (décantation, hydrocyclones, filtres mécaniques à tambour rotatif, filtration gravitaire) ne donnent par entière satisfaction. Par contre, la filtration sur membranes permet d'arrêter en totalité les particules en suspension et les bactéries.

On montre d'abord que les teneurs en M.E.S. et leurs répartitions granulométriques mesurées sur des échantillons prélevés en bassins d'aquaculture varient avec la taille des poissons et l'heure du prélèvement et on met en évidence la présence en nombre important de particules submicroniques.

Différents essais de filtration sur membranes sont ensuite présentés :

- d'une part, avec des membranes d'ultrafiltration capillaires à peau interne utilisées en potabilisation des eaux : on examine l'influence des paramètres hydrodynamiques (pression transmembranaire, vitesse de recirculation) afin de rechercher les conditions optimales de fonctionnement. Le flux de perméat ne dépasse pas dans le meilleur des cas 100 l.h^-1.m^-2.

- d'autre part, avec des membranes de microfiltration organiques planes en fluorure de polyvinylidène (PVDF) et tubulaires en céramique. Les flux obtenus avec les membranes organiques sont de l'ordre de 250 l.h^-1.m^-2

Dans tous les cas, la rétention des M.E.S. est totale.

Cependant l'estimation de l'investissement et des coûts de fonctionnement pour une pisciculture en circuit fermé de taille industrielle conduit à des prix trop élevés pour que l'utilisation des membranes dans ce domaine soit à ce jour économiquement envisageable.

EN :

A problem confronting semi-closed circuit aquaculture is the need for continuous elimination of suspended matter (SM) and nitrogenous substances. Conventional processes used to retain SM (settling tanks, hydrocyclones, rotating-drum mechanical filters, gravity filtration) are not entirely satisfactory. However, membrane filtration has recently been shown to allow removal of suspended particles and bacteria. The present study evaluates the performance of different ultrafiltration and microfiltration membranes for water processing in a semi-closed aquaculture system. A brief economic analysis of treatment costs is proposed based on the results.

The marine aquafarm studied produces about 5 tons of turbot per year with a plant volume of about 100 m³. The water processing line is fitted with a rotating-drum mechanical filter that stops the largest particles and ejects 1 m³ h^-1 of loaded water into the surrounding environment. Another 2 m³ h^-1 are cleared out by overflowing the pumping pit. These volumes are renewed at a rate of 3% per hour by pumping saltwater from an underground source. Crossflow filtration was performed on rejections from both the mechanical filter and pumping pit overflow. SM contents and granulometric distributions determined by laser diffractometry were found to vary with sample source and withdrawal time, and size of fish in the pens. A comparison of granulometric distributions in volume percent and numerical percent underscores the presence of a great number (> 98 %) of submicron particles.

To limit the risk of mechanical-pore fouling due to blockage by particles, organic membranes in the form of internal-skin capillaries (pore diameters of about 10 to 20 nm) were initially employed. These membranes, used in drinking water production, are relatively inexpensive. The experimental device was fitted with an interchangeable volumetric pump (with gears or monoscrew). Adjustable parameters were transmembrane pressure and circulation velocity within the module. Analysis of the influence of these hydrodynamic parameters revealed that pressures higher than 1 bar were unnecessary, as beyond this point permeate flux no longer increased. Optimal flux did not exceed 100 L h^-1 m^-2 with the gear pump. Replacing the latter with a monoscrew pump improved permeate flux up to 70 %.

Tests were also performed with flat microfiltration organic membranes of polyvinylidene fluoride (PVDF) with pore diameters ranging from 0.1 to 8 µm. The flux obtained with these membranes was roughly 250 L h^-1 m^-2 and presented little variation with varying pore diameter. Comparative tests carried out on tubular membranes showed lower fluxes than those obtained with organic membranes which, considering their much higher cost, makes them less attractive in this context.

The use of membranes in aquafarming is without precedent. An economic analysis of the practice was carried out based on financial assessments of processing of surface waters into drinking water, for which outputs to be treated and SM contents were of the same order of magnitude. With operating costs from 0.35 to 0.95 FF per cubic meter of filtered water, expected investment for a fishfarm producing 100 tons of fish a year is currently 3 to 4 times too great to consider economically profitable the use of membranes for water treatment in closed-circuit aquafarming.

FR :

Une étude microbiologique et chimique a été menée pendant un an sur les eaux de sources et de puits de Yaoundé (Cameroun). Les analyses microbiologiques ont été faites suivant la technique des membranes filtrantes, et les analyses chimiques, suivant les techniques analytiques usuelles. Les abondances maximales mensuelles de Pseudomonas aeruginosa et de Aeromonas hydrophila varient respectivement de 1 à 22x10³ UFC.100 ml^-1 d'eau, et de 1 à 7,8x10³ UFC.100 ml^-1. Ces abondances bactériennes subissent d'amples fluctuations spatio-temporelles. Les eaux de sources et de puits analysées sont faiblement bicarbonatées, douces et présentent une minéralisation faible à moyenne. Le pH varie de 3 à 5 et les concentrations en CO₂ dissous, de 300 à 532 mg.l^-1. Un bon nombre des caractéristiques chimiques de ces eaux est relativement stable au cours du temps, en dépit des fluctuations spatiales apparentes. Le degré de corrélation entre les paramètres chimiques et la dynamique d'abondance des bactéries isolées, est hétérogène. Cela est nettement plus marqué dans les eaux de puits, en raison des conséquences de l'exploitation humaine de ces eaux, à l'origine d'apports allochtones divers.

EN :

bacteriological and chemical study was carried out on the waters of four springs and six wells over 13 months and 12 months, respectively, in the town of Yaounde and its environs in Cameroon. For these groundwater points the study analysed the importance of pH, electrical conductivity, dissolved oxygen and carbon dioxide, and some ions such as sodium, potassium, calcium, magnesium, chloride and bicarbonate, for communities of Pseudomonas aeruginosa and Aeromonas hydrophila. The two bacteria P. aeruginosa and A. hydrophila are considered opportunistic pathogens. Different groundwater points were chosen on the basis of their spatial distribution and the size of the human population using them. Microbiological analyses were carried out by membrane filtration and chemical analyses were performed using standard analytical techniques.

It has been noted that the monthly maximum abundance of P. aeruginosa and A. hydrophila varies from 1 to 22 x 10³ CFU per 100 ml and from 1 to 7.8 x 10³ CFU per 100 ml of water, respectively. These bacteria are sometimes rare and their abundance undergoes spatio-temporal fluctuations. The studied waters are slightly bicarbonate and soft, with a low to average level of mineralisation. The pH varied from 3 to 5 and the concentration of dissolved CO₂ from 300 to 532 mg/l. Most chemical characteristics were relatively stable with time, but exhibited apparent spatial fluctuations. The level of correlation between the chemical parameters and the abundance dynamics of P. aeruginosa and A. hydrophila is heterogeneous. Depending on the origin (well or spring) of water samples and minimising their spatial variations, samples were grouped into single biotopes (either mother well or mother spring) which were separated into compartments. A correlation test was then carried out using data obtained from 52 monthly-average samplings in the four compartments of mother-spring water, and using data obtained from 72 monthly-average samplings in the six compartments of mother-well water. From this it emerged (P < 0.001) that in springs increases in pH, electrical conductivity, and in concentrations of chloride, sodium, potassium, calcium, magnesium and dissolved oxygen favoured the development of P. aeruginosa and A. hydrophila. High concentrations of dissolved carbon dioxide reduced (P < 0.001) the abundance of these bacteria. In wells, the ecology of P. aeruginosa and A. hydrophila is relatively unstable in comparison with springs. The higher instability in well water is ascribed to an increase in the number of confounding factors, which make wells appear more vulnerable than springs. This leads to the multiplication of asymmetrical interaction networks affecting bacterial population dynamics.

FR :

Des études récentes ont montré que, lors du rejet d'eaux usées dans une rivière, la quantité de biomasse bactérienne hétérotrophe amenée par les effluents influence considérablement la cinétique de biodégradation de la matière organique dans la rivière et donc les caractéristiques du déficit d'oxygène généralement observé dans le milieu naturel en aval du rejet. La mesure de la biomasse bactérienne contenue dans un rejet domestique est donc nécessaire afin de bien comprendre la cinétique de biodégradation. Cette biomasse peut être estimée en microscopie à épifluorescence après coloration des cellules bactériennes par un fluorochrome. Cette technique appliquée aux eaux usées est néanmoins difficile et fastidieuse. Dans cette étude, une méthode alternative à l'estimation de la biomasse bactérienne dans les eaux usées a été testée ; elle consiste à mesurer l'activité exoprotéolytique potentielle (AEP) des bactéries. Nous avons montré qu'il existait, dans les eaux usées, une corrélation significative entre l'AEP et la biomasse bactérienne estimée en microscopie à épifluorescence ce qui permet d'utiliser l'AEP pour estimer facilement et rapidement la biomasse bactérienne dans ce type d'échantillon. Comme exemple d'application, des mesures d'AEP nous ont permis d'étudier l'impact de divers types de traitement dans plusieurs stations d'épuration sur la biomasse bactérienne hétérotrophe des effluents urbains. Sur base de ces mesures, les charges spécifiques en biomasse bactérienne (charge par habitant et par jour) des eaux brutes et traitées ont pu être calculées.

EN :

Recent studies have shown that when wastewaters are discharged into a river, heterotrophic bacterial biomass in the effluent exerts considerable influence on the biodegradation kinetics of organic matter and provokes the oxygen deficits often observed downstream from outfalls. Quantification of bacterial biomass in wastewaters is required for a good understanding the biodegradation kinetics. Bacterial biomass can be estimated by epifluorescence microscopy after staining cells with fluorochrome. Applying this technique to wastewater samples, however, is painstaking and difficult. In this study, an alternative method for estimating bacterial biomass in wastewaters was tested; it consists of measuring the potential exoproteolytic activity (PEA) of bacteria. Previous studies on other types of aquatic systems have shown that bacterial biomass can be estimated from PEA. After optimising the PEA procedure for use with wastewater, we found a significant correlation between PEA and the bacterial biomass estimated by epifluorescence microscopy. This opens the way for PEA measurements for rapid and easy estimates of bacterial biomass in wastewaters. As an example of the application of this technique, the impact of wastewater treatment on bacterial biomass in effluent was investigated at various treatment plants. On the basis of our procedure, specific loads of bacterial biomass (load per inhabitant and per day) were calculated for raw and treated wastewaters.

FR :

L'objectif de ce travail était d'apprécier l'intérêt ou la redondance de plusieurs bioessais à travers leur application à des échantillons particulièrement complexes. Huit bioessais, plus ou moins couramment utilisés en écotoxicologie, ont donc été testés sur une série de 14 échantillons liquides dont 11 étaient représentés par des effluents de lixiviation de déchets urbains et industriels. L'analyse statistique des résultats montre que les bioessais mis en œuvre présentent des réponses et sensibilités très différentes. Ils sont donc tous informatifs et une évaluation écotoxicologique globale devrait donc tous les prendre en compte, ce qui n'est pas possible dans la réalité. Il apparaît cependant que, pour une surveillances de routine de milieux complexes, le test Microtox® est globalement sensible, facile à mettre en œuvre et applicable à de grandes séries d'échantillons. Il est néanmoins peu représentatif des milieux naturels et d'autres bioessais ne peuvent que le compléter utilement, en particulier celui visant à déterminer le caractère mutagène du milieu étudié.

EN :

Eight ecotoxicological bioassays were evaluated using 14 highly complex liquid samples, including 11 effluents from urban waste and industrial waste lixiviation. Statistical analysis of results revealed that the bioassays have widely varying sensitivities. All were informative-ideally, general ecotoxicological evaluation should spare none. However for routine evaluation of complex liquid mixtures, the Microtox® test is sufficiently sensitive and easy to use and accommodates a large range of samples. Nevertheless Microtox® does not accurately represent natural media and other bioassays, particularly mutagenic tests, are useful complements.

FR :

Les eaux de l'Oued Hassar (Maroc) sont caractérisées par une forte minéralisation. La conductivité, variant entre 3470 et 7210 µS.cm^-1, est due essentiellement aux fortes teneurs en chlorures. L'examen de l'Indice de Pollution Organique (IPO) montre que l'effluent d'eaux usées provenant de la ville de Mediouna constitue la principale source de pollution des eaux de l'Oued Hassar.

Concernant l'étude des diatomées épilithiques, 61 espèces et variétés de diatomées ont été inventoriées. Ces taxons sont liés particulièrement à la salinité (Pleurosira laevis, Fragilaria pulchella,…) et à la charge organique (Nitzschia palea, Gomphonema parvulum,…).

Quatre indices diatomiques ont été testés pour l'évaluation de la qualité des eaux de l'Oued Hassar : l'Indice Diatomique Pratique (IDP), l'Indice Diatomique Générique (IDG), l'Indice de Sládecek (SLA) et l'Indice Diatomique Trophique (IDT). La variation spatio-temporelle de ces indices reflète bien la dégradation de la qualité de ce cours d'eau, particulièrement en aval du rejet de la ville de Mediouna (IDP=3,86 ± 1,84 ; IDG=6,05 ± 1,95 ; SLA=4,35 ± 1,93 ; IDT=5,80 ± 1,31). Parmi ces indices testés, seul l'IDP présente une corrélation significative (0,63 ; α<5%) avec l'IPO (indice de référence) tandis que l'IDG, le SLA et l'IDT montrent des corrélations non significatives avec cet indice (0,38 ; 0,27 et -0,11 respectivement).

EN :

The water quality in Moroccan rivers and streams becomes more and more injured because of the important amount of pollutants discharged in these aquatic ecosystems without any treatment in the most cases. A physical, chemical and microbiological analyses are regularly done by public institutions (office national des eaux potables, direction de l'hydrolique). However, these analyses become insufficient facing the big diversity of chemicals and the intermittence of pollution sources at the time. So, methods using a biotic indicators as a complement of the chemical analyses become a requirement in the water quality assessment. In the present study, besides the physical and chemical analyses, we have tested the use of benthic diatoms as indicators of the oued Hassar water quality and diatom indices as a tool for the evaluation of this quality.

Oued Hassar is a canalized stream situated on the valley of oued Mellah river. It drains a part of the Berrechid groundwater as sources. Its main source (station SC) is situated to 3 km in the Northeast of Mediouna (155 m). Along this water course, in sections of various degrees of water pollution, six sites (SC, SB, HP, OM, SH and CA) were sampled on four different dates (august 1997 and january, june and august 1998). The station OM belongs to oued Mouileh stream which is a unique tributary of oued Hassar. The main source of pollution; the wastewater effluent of Médiouna agglomeration entered the stream between The oued Hassar source and the station SB (Sidi Brahim). This region has a Mediterranean semi-aride bioclimate. The influence of the ocean results in a summer temperatures generally varying between 25 and 30°C and winter temperatures varying between 12 and 19°C.

The physical and chemical analysis data of the oued Hassar waters show particularly elevated values of water conductivity essentially due to the high contents in chlorides. These values were understood between a minimal value recorded at station SH in August 1998 (3470 mS.cm^-1) and a maximal value signalled at station OM in January of the same year (7210 mS.cm^-1). Concentrations in sulphates vary according to stations. The most elevated values were recorded in january 1998 at stations SC and OM (231 and 218 mg.l^-1 respectively). The hardness values are very important especially at the source of oued Hassar (SC) where the concentration reaches 25,2 meq.l^-1. This excessive mineralization of waters, particularly at stations SC and OM, is essentially due to the Permo-triasique saliferous sedimentary layers which crop out in the area.

The oued Hassar stream receives, to 100 ms downstream the station SC, the wastewater effluent of Médiouna agglomeration, highly loaded with organic matters, ammonia, orthophosphates and suspended matters. The impact of this effluent results on the elevated values of these parameters downstream the wastewater input (SB). On January 1998 (rainy period), the important contents in orthophosphates and sulphates recorded at all stations were essentially due to the high amounts of superphosphates and ammonia sulphates used as amendment of the cultivated lands situated a long the sides of oued Hassar stream.

Outside of this period, contents in orthophosphates and ammonia at stations SC and OM are very low. However, the elevated content in nitrates at the oued Hassar's source (maximal value: 93,3 mg.l^-1) gives an idea on the deterioration of the groundwater quality.

To put in evidence the evolution of the global organic pollution of waters, the "Indice de Pollution Organique" (IPO) have been calculated. The exam of this index shows that the wastewater effluent constitutes the main source of pollution. Indeed, water quality goes from a mean pollution state (3,88 ± 0,14) at the oued Hassar's source to a very strong pollution state (1,96 ± 0,31) at the station SB. Downstream the swampy zone, at the station HP, the water quality of oued Hassar improves remarkably (3,50 ± 0,74) due to the auto-purification phenomenon and the dilution of these waters by those coming from the secondary sources situated between stations SB and HP. At the stations SH and CA, the IPO mean values (3,50 ± 0,79 and 3,75 ± 0,41 respectively) show that the water quality remains rather identical as at the station HP.

Concerning the survey of epilithic diatoms, 61 species and varieties of diatoms have been inventoried. These taxa are bound especially to the salinity (Amphora coffeaeformis, Fragilaria fasciculata, Fragilaria pulchella, Pleurosira laevis,…) and to the organic load (Gomphonema parvulum, Navicula subminuscula, Nitzschia palea,…). In the most collected samples, particularly at the stations SB, SH and CA, diatom communities are essentially dominated by N. palea and the complex N. inconspicua / N. frustulum. At the station SB, the relative abundance of N. palea reaches 71% in august 1997 and august 1998, and goes beyond 45% in January and june 1998. giving evidence of the important organic pollution. this species is substituted, at stations SH and CA, essentially by Navicula subminuscula, N. veneta and Nitzschia inconspicua / N. frustulum complex. Diatom communities at the stations SC, HP and OM don't show an exclusive dominance of one or two species, but several species dominate and form a more varied communities.

Four diatom indices have been tested to assess the water quality in Oued Hassar (Morocco) : the "Indice Diatomique Pratique" (IDP), the "Indice Diatomique Générique" (IDG), the Sládecek Index (SLA) and the Trophic Diatom Index (TDI). These indices are different by the assigned indicative values and pollution sensitivity coefficients relative to every species and by the taxonomic list integrated in the calculation of every index. The spatio-temporal variation of these indices reflects clearly the deterioration of water quality downstream the wastewater input of Mediouna agglomeration (IDP=3,86 ± 2,30; IDG=6,05 ± 2,31; SLA=4,35 ± 2,62; IDT=5,80 ± 1,85) and its recovery after the swampy zone at the station HP (IDP=9,37 ± 2,35; IDG=11,83 ± 1,77 ; SLA=7,19 ± 3,05 ; IDT=6,76 ± 2,48). Among these indices, only the IDP presents a significant correlation (0,63 ; α<5%) with the IPO (reference index). While the IDG, the SLA and the TDI don't show a significant correlation with this index (0,38 ; 0,27 and -0,11 respectively). The IDT shows nearly the same water quality in the different stations and rather inform on the high level of eutrophication in these water courses.

The physical and chemical analyses, the composition of diatom communities as well as the relative results of diatom indices, show that the oued Hassar stream is submitted to an important pollution, particularly downstream the wastewater input. Among the tested indices, the IDP appears more adapted to the assessment of water quality in oued Hassar. However, more work is required in order to finalize a new diatom index which is more adapted to characteristics of the Moroccan water courses.

FR :

Le modèle de l'IAWQ du processus de boues activées représente les mécanismes endogènes de la biomasse nitrifiante par le décès des micro-organismes (équation d'ordre 1 par rapport à la biomasse). La constante de décès, ou taux de mortalité b_A, est aujourd'hui encore mal connue, et en particulier les facteurs influants sur sa valeur. De récentes études ont montré que la prédation par la microfaune pourrait être un facteur déterminant sur la valeur de b_A. Cette étude se propose donc de quantifier l'effet de la prédation sur la valeur de b_A. Deux réacteurs maintenus sans alimentation en substrat ont été caractérisés en parallèle: l'un a reçu une dose d'antibiotique spécifique aux eucaryotes (cycloheximide) afin de diminuer la quantité d'organismes de la microfaune, alors que l'autre n'a reçu aucun antibiotique (témoin). Les résultats obtenus montrent que le cycloheximide inhibe la plupart des organismes de la microfaune sauf les amibes; celles-ci semblent plutôt stimulées par cet antibiotique. En ce qui concerne la nitrification, un ralentissement de la production de nitrate dans le réacteur traité à l'antibiotique est observé à partir du sixième jour. Cette diminution de production de nitrate est probablement causée par une réduction de l'azote nitrifiable (qui est mobilisé par les amibes) couplée à une prédation des organismes nitrifiants par les amibes. D'ailleurs, l'augmentation de la prédation par les amibes à partir du jour 6 a diminué l'activité nitrifiante également mesurée par respirométrie (rO₂ Nmax). Cette diminution du taux de respiration indique une augmentation du taux de mortalité (bA) des organismes nitrifiants. En effet, la valeur du taux de mortalité mesurée dans le réacteur témoin est de 0.08 d^-1 alors que selon la microfaune présente dans le réacteur inhibé au cycloheximide, la valeur de ce taux de mortalité a varié entre 0.05 d^-1 et 0.15 d^-1.

EN :

Designing biological wastewater treatment plants with the aid of the model developed by the IAWQ requires the knowledge of biological kinetic parameters. For nitrifying activated sludge, these parameters are related to nitrifying bacteria: maximum autotrophic growth rate µ_Amax, yield coefficient YA and the autotrophic decay rate b_A. Although variables influencing µ_Amax and Y_A values are well known, this is not the case for bA. MARTINAGE and PAUL (2000) have recently shown that the bA value is strongly influenced by the influent quality, leading to the assumption that influent quality has a strong effect on microfauna composition, and consequently on the grazing rate of microfauna on nitrifying bacteria. In fixed-film processes, protozoan grazing reduces the bacterial population considerably (NATUSCKA and WELANDER, 1994). However, although many data are available concerning the grazing rates of different protozoa, the effect of microfaunal grazing on nitrification is still a matter of debate (RATSAK et al., 1994) and its effect on the b_A value is still unknown. These two topics are investigated here.

Nitrifying activated sludges were grown in two identical batch reactors, but in one, cycloheximide was added to inhibit eucaryotic growth (MAURINES CARBONEILL et al., 1998). Microfauna organism numbers were quantified in both reactors by microscopic observations of flagellated protozoa (>8 µm), amoebae, ciliates, rotifers and higher invertebrates (Fig. 3). Microbial counts were then correlated with the b_A value. The latter was determined using the procedure proposed by SALZER (1992) which consists of characterising the time behaviour of the maximum nitrification rate measured by respirometry of activated sludge under substrate starvation. Under these conditions bacteria die and organic nitrogen is released into the bulk phase. This nitrogen is ammonified, and nitrifying bacteria use this substrate to produce nitrate, and then autotrophic bacterial growth occurs. This method takes this growth into account by characterising nitrate production during the experiment (Fig. 2).

The effect of cycloheximide on nitrification was first determined to make sure that this compound is not inhibitory toward nitrifiers. Results obtained (Table 1) show that cycloheximide was not inhibitory toward nitrate production or the maximum nitrification oxygen uptake rate (rO₂ N) after 4 hours of contact with nitrifying biomass. Cycloheximide addition in the activated sludge had an important impact on rotifers and flagellates but no effect on ciliates; it also seemed to stimulate amoebae growth. In both reactors, flagellates were mainly Peranama, attached ciliates were mainly Opercularia and Epistylis and a few Vorticella. Free ciliates like Aspidisca and Euplotes were found in both reactors.

Variation with time of the abundance of microfauna organisms is shown in Figures 4 and 5 for both reactors. In the reference reactor the number of microfauna organisms decreased with time (Fig. 4) probably due to substrate starvation. Microfauna composition remained however diversified. For the inhibited reactor (Fig. 5), three periods were observed. During period I, the microfauna was mainly composed of ciliates and the number of microfaunal organisms decreased rapidly. During period II, an important growth of amoebae was observed. Cycloheximide was then added during this period to reduce their number. This growth of amoebae seems to be caused by the resistance of these micro-organisms toward inhibiting compounds (SRIKANTH et BERK, 1993). During period III, the number of microfaunal organisms was lower than during period II, and microfauna was mainly composed of ciliates.

Nitrate concentration behaviour, necessary for bA calculation, is shown on Figure 6. In the reference reactor, nitrate concentrations varied linearly. For the inhibited reactor, the linear pattern was not observed during period II. This result was probably caused by an important nitrogen assimilation need of amoebae (ELDRIGE and JACKSON, 1993). Because organic nitrogen released by bacterial decay is consumed by amoebae assimilation, less nitrogen is available for the ammonification process and therefore for nitrification. Ammonia concentrations remained below 0.2 mg N·l^-1 during all the experiment for both reactors. When amoebae disappeared from the inhibited reactor (period III) nitrate concentration varied linearly again.

Variations of the maximum nitrification oxygen uptake rate (rO₂ N_max) with time are presented in Figure 7 (A&B) for both reactors. Two curves are plotted on each figure. Empty squares represent the measured rO₂ N and black points represent the maximum nitrification rate that would have been measured if no growth on ammonification products had occurred. For the reference reactor (Fig. 7A), a value for b_A of 0.08 d^-1 can be calculated and can be considered constant for a constant microfauna composition.

Three b_A values can be estimated for the reactor inhibited with cycloheximide (Fig. 7B), corresponding to the three periods observed for microfauna composition. During period I, the bA value is 0.05 d^-1 : a decrease in the microfaunal organism numbers implies a decrease of the b_A value. During period II, when a development of amoebae is observed, the bA value increases and reaches 0.15 d^-1. During period III with reduced grazing, the b_A value is 0.13 d-1. Since during periods I and III the microfauna is mainly composed of ciliates, this difference between b_A values is likely due to the observed difference in floc size between periods I and III.

The results obtained during this study tend to prove (1) that the use of cycloheximide reduces microfaunal populations but can lead to a development of amoebae, and (2) that microfauna grazing seems to have an influence on the b_A value, which can vary from 0.05 to 0.15 d^-1 depending on microfaunal composition and abundance.