Thesis (PhD)--Stellenbosch University, 2012. / ENGLISH ABSTRACT: The accuracy of methods for the enumeration of coliforms and Escherichia coli present in river water intended for the irrigation of fresh produce has been critically evaluated to determine whether the results of the traditional method were reliable in indicating faecal pollution. The potential of rapid alternative methods were also explored. Baseline monitoring of the Berg River showed the presence of potential pathogens such as Salmonella, and also that E. coli levels exceeded international guidelines for the safe irrigation of minimally processed foods (MPFs) in 20.5% of cases, which indicated faecal intrusion. An exploratory study into the use of microbiological and physico-chemical parameters in predicting E. coli numbers, as a rapid alternative to direct enumeration, was conducted. These measurements, neither individually nor in combination, could accurately predict the E. coli numbers. The rapid method Colilert-18 was compared against multiple tube fermentation (MTF) for the enumeration of coliforms and E. coli. Spearman rank correlation coefficients showed that Colilert-18 had acceptable (r2=0.69) and fair (r2=0.74) correlations with MTF for coliform and E. coli enumeration, respectively. Bland and Altman statistics were used to determine pollution influence, and Colilert-18 showed increasing disagreement with MTF at very high concentrations of coliforms and E. coli. Bacterial isolates obtained from MTF reactions were identified using biochemical and mass spectrometry methods. These identifications revealed that the greatest contributors to inaccurate coliform enumeration by MTF were false negative coliforms which fail to produce gas from lactose. Numerical biochemical data suggested that these isolates may be able to use other carbohydrates preferentially over lactose. Inaccurate E. coli enumeration was caused by E. coli strains which could not utilise lactose or 4-methylumbelliferyl- -D-glucuronide (MUG), as well as non-E. coli isolates which were able to hydrolyse MUG. The method of transfer of bacteria between MTF media was also identified as problematic for accuracy. Monoplex polymerase chain reaction (PCR) differentiation of MTF isolates showed that detection of the uidA gene showed the greatest accuracy in the detection of E. coli, while the multiplex PCR protocol for detecting diarrheagenic E. coli pathotypes identified one strain of enteroaggregative E. coli (EAEC). A qualitative methodological risk classification was used, in combination with the individual reactions of MTF isolates, to elucidate their contribution to enumeration inaccuracy and to evaluate the effect of MUG and Levine-eosin methylene blue (L-EMB) agar. The classification indicated that inaccurate enumeration of E. coli was more problematic than that of coliforms, but revealed that the exclusion of MUG from MTF may increase the accuracy of E. coli enumeration. The omission of L-EMB would have very little effect on E. coli enumeration accuracy. This work confirmed that MTF is fairly reliable in the enumeration of coliforms and E. coli. Inaccuracies are primarily attributable to atypical organisms which are considered to make up a small proportion of the total bacterial population. Colilert-18 was shown to be an acceptably accurate alternative, and its rapid production of results can be highly advantageous in the monitoring of irrigation water used for MPFs. / AFRIKAANSE OPSOMMING: Die akkuraatheid van metodes vir die telling van kolivorme en Escherichia coli in rivier water, gebruik vir besproeiing van vars produkte, is krities geëvalueer om vas te stel of die resultate van tradisionele metodes betroubaar was in die aanduiding van fekale besoedeling. Die potensiaal van snelle alternatiewe metodes is ook ondersoek. Basislyn monitering van die Berg River het aangedui dat potensiële patogene soos Salmonella teenwoordig is, en dat E. coli vlakke internasionale riglyne vir die veilige besproeiing van minimaal geprosesseerde voedsels (MGVs) oorskry het in 20.5% van gevalle, wat dui op fekale besmetting. ‘n Ondersoekende studie van die bruikbaarheid van mikrobiologiese en fisieschemiese parameters in die voorspelling van E. coli getalle, as snelle alternatief tot direkte telling, is uitgevoer. Hierdie metings kon nie, individueel of in kombinasie, akkurate voorspellings van E. coli getalle maak nie. Die snelle metode Colilert-18 is vergelyk met veelvoudige buis fermentasie (VBF) in die telling van kolivorme en E. coli. Spearman rang korrelasie koëffisiënte het aangetoon dat Colilert- 18 aanvaarbare (r2=0.69) en goeie (r2=0.74) korrelasies met VBF gehad het vir kolivorm en E. coli tellings, respektiewelik. Bland en Altman statistiek is gebruik om die invloed van besoedeling te bepaal, en Colilert-18 het afnemende ooreenstemming met VBF getoon by baie hoë kolivorm en E. coli konsentrasies. Bakteriële isolate verkry vanaf VBF reaksies is geïdentifiseer met behulp van biochemiese en massa spektrometrie metodes. Hierdie identifikasies het getoon dat vals-negatiewe kolivorme, wat nie gas vanaf laktose kan produseer nie, die grootste bydraende faktor is in onakkurate kolivorm telling deur VBF. Biochemiese data het voorgestel dat hierdie isolate moontlik ander koolhidrate by voorkeur bo laktose gebruik. Onakkurate E. coli tellings is veroorsaak deur E. coli isolate wat nie laktose of 4-metielumbelliferiel- -D-glukuronied (MUG) kon verbruik nie, sowel as nie-E. coli isolate wat wel MUG kon hidroliseer. Die oordrag-metode van bakterieë tussen VBF media is ook geïdentifiseer as problematies. Monopleks polimerase ketting reaksie (PKR) onderskeiding van VBF isolate het aangedui dat opsporing van die uidA geen die grootste akkuraatheid vir die opsporing van E. coli het, terwyl die multipleks PKR protokol vir die opsporing van diarree-veroorsakende E. coli patotipes een stam van entero-aggregerende E. coli (EAEC) geïdentifiseer het. ‘n Kwalitatiewe metodologiese risiko klassifikasie is gebruik, in kombinasie met die individuele reaksies van VBF isolate, om hul bydrae tot telling onakkuraatheid vas te stel. Die effek van MUG en Levine-eosien metileenblou (L-EMB) agar is ook geëvalueer. Daar is bevind dat die onakkurate telling van E. coli meer problematies is as dié van kolivorme, maar ook dat die uitlating van MUG by VBF die akkuraatheid van E. coli tellings kan verhoog. Die uitlating van LEMB agar sal ‘n ignoreerbare effek hê op E. coli telling akkuraatheid. Hierdie werk het bevestig dat VBF akkuraat is in die telling van kolivorme en E. coli. Onakkuraathede word primêr toegeskryf aan atipiese organismes, wat beskou word as ‘n klein proporsie van die totale bakteriële bevolking. Colilert-18 is ‘n aanvaarbaar akkurate alternatief, en die metode se produksie van snelle resultate kan hoogs voordelig wees in die monitering van MGVs besproeiingswater.

No description available.

As fontes hídricas disponíveis para o consumo humano vêm sendo comprometidas. Para resolver este problema, várias técnicas de detecção de contaminação fecal foram desenvolvidas. Em 1982, Manja e colegas desenvolveram método H2S que é simples, rápida e de baixo custo e detecta bactérias produtoras de H2S e, assim, a contaminação fecal. Neste trabalho foi analisada a eficácia de detecção de micro-organismos produtores de H2S frente a adição de L-cistina (125mg/L e 250mg/L) e desoxicolato de sódio (DS) (0,1% e 0,3%) e na presença de bactérias não produtoras de H2S para verificar se a presença destas bactérias interferem na detecção dos isolados H2S+. Assim, comparou-se o teste H2S com a membrana filtrante e o Colilert®. Os resultados deste estudo indicam que o meio H2S com adição de 0,3% de desoxicolato de sódio se mostrou mais rápido e sensível. Quando comparado com outras metodologias clássicas, o meio com 0,3% apresentou uma ligeira queda na sensibilidade mas o método H2S se mostrou mais sensível que o Colilert. / Water supply for human consumption have been compromised. Several detection methods for fecal contamination have been developed to solve this problem. Manja and co-workers (1982) developed a simple, fast and low-cost method for fecal contamination based on detection of sulfate-reducing bacteria, the H2S method. This work aimed to analyse the detection efficiency of the H2S method under different conditions: with L-cystine (125mg/L e 250mg/L) and sodium deoxycholate (0.1% e 0.3%). Also, non-sulfate-reducing bacteria interference were evaluated. Comparison tests were made through membrane filtration and Colilert®. Our results indicate a faster and more sensible for the 0.3% sodium deoxycholate condition. Compared to other classic methodologies, the 0.3% sodium deoxycholate condition slightly decrease the sensibility. However the H2S method was more sensitive than the Colilert one.

As fontes hídricas disponíveis para o consumo humano vêm sendo comprometidas. Para resolver este problema, várias técnicas de detecção de contaminação fecal foram desenvolvidas. Em 1982, Manja e colegas desenvolveram método H2S que é simples, rápida e de baixo custo e detecta bactérias produtoras de H2S e, assim, a contaminação fecal. Neste trabalho foi analisada a eficácia de detecção de micro-organismos produtores de H2S frente a adição de L-cistina (125mg/L e 250mg/L) e desoxicolato de sódio (DS) (0,1% e 0,3%) e na presença de bactérias não produtoras de H2S para verificar se a presença destas bactérias interferem na detecção dos isolados H2S+. Assim, comparou-se o teste H2S com a membrana filtrante e o Colilert®. Os resultados deste estudo indicam que o meio H2S com adição de 0,3% de desoxicolato de sódio se mostrou mais rápido e sensível. Quando comparado com outras metodologias clássicas, o meio com 0,3% apresentou uma ligeira queda na sensibilidade mas o método H2S se mostrou mais sensível que o Colilert. / Water supply for human consumption have been compromised. Several detection methods for fecal contamination have been developed to solve this problem. Manja and co-workers (1982) developed a simple, fast and low-cost method for fecal contamination based on detection of sulfate-reducing bacteria, the H2S method. This work aimed to analyse the detection efficiency of the H2S method under different conditions: with L-cystine (125mg/L e 250mg/L) and sodium deoxycholate (0.1% e 0.3%). Also, non-sulfate-reducing bacteria interference were evaluated. Comparison tests were made through membrane filtration and Colilert®. Our results indicate a faster and more sensible for the 0.3% sodium deoxycholate condition. Compared to other classic methodologies, the 0.3% sodium deoxycholate condition slightly decrease the sensibility. However the H2S method was more sensitive than the Colilert one.

MSc (Microbiology) / Department of Microbiology / See the attached abstract below