Miljøprojekt, 1062 – Kildeopsporing ved fænotypisk karakterisering af enterokokker

Source Tracking by Phenotypic Characterisation of Enterococci

Summary and conclusions

The Proposal for Directive of the European Parliament and of the Council concerning the quality of bathing water (COM/2002/0581 final) put more focus on the improvement of the hygienic bathing water quality than previously. In the proposal, legally binding limit values are proposed for the hygienic indicators E. coli and Enterococci. To improve the hygienic water quality on non-complying locations, it is necessary to know the source of the contamination. Therefore there is a need to test and apply methods to track the origin of the hygienic indicators.

PhPlate Rapid is a screening method for microbial source tracking, which enables you to characterise bacteria on the basis of their phenotypic properties. The phenotypic properties of a bacterium can be defined as "the visible physical characteristics or behaviours" of the bacterium, e.g. its growth rates on different substrates under specified conditions. In order to use PhPlate Rapid, bacteria are isolated from a contaminated bathing water location and from the potential sources of the contamination and phenotyped. A subsequent statistical analysis can then show which of the potential sources is most probably causing the contamination.

It was the aim of this project to evaluate PhPlate Rapid for Enterococci as a tool to track the sources of hygienic contamination in bathing water, and specifically to track the source of contamination in the eastern part of Lake Vandet (Vandet Sø Østende).

Lake Vandet was chosen because the concentration of E. coli exceeds the accepted limits every summer, and because the Municipality of Thisted has examined the area within the last few years and identified the most likely sources of contamination.

Samples were taken from the bathing water location and from the potential sources of contamination. From these samples, Enterococci were isolated by cultivation on m-Enterococcus agar and subsequent verification on bile-aesculine-azide agar and characterised phenotypically by PhPlate Rapid. It was the aim to achieve 24 isolates from each sample except from the faeces samples, for which 8 isolates were the aim. In the project, a total of 238 Enterococci were isolated.

In the PhPlate Rapid method, the ability of Enterococci to grow on different relevant substrates is examined. Eleven dried substrates have been added to each row of an 8 times 12 well microtitre tray. At the start of the analysis, enterococ-medium containing bromothymol blue as an indicator of growth is added to all the wells of the PhPlate microtitre tray. An Enterococ-colony is suspended in the medium of the first well, which does not contain a dried substrate. From the first well, the Enterococ-suspension is pipetted into the remaining 11 wells. This procedure is repeated for the remaining 7 rows of the PhPlate microtitre tray. The PhPlate is then incubated and read on a flatbed-scanner after 16, 40 and 64 hours. Wells developing a yellow colour are positive.

Samples were taken from the bathing water location at the eastern part of Lake Vandet and from the potential sources of contamination on 21 August and 10 September 2003. The potential sources of contamination were a sewage infiltration plant, a drainage canal, a summer camp facility, pig slurry, cattle faeces, sedimented material and gulls' faeces.

Prior to the analysis of the samples from the eastern part of Lake Vandet, a number of preliminary analyses were carried out on control strains. The results were acceptable for two of three control strains whereas the third control strain had a variable phenotype and showed a higher variation than expected. Analysis of blinds (sterile water) showed that unintended contamination might occur leading to unwanted variation. The analysis of control strains and double analyses showed that the variation as a whole was acceptable. Based on the observed variation, the limit, above which two isolates are considered to be of the same phenotype, was determined to be 0.965.

The 2003 results of the obligatory analyses of the bathing water from the eastern part of Lake Vandet failed to meet the hygienic standards. The results of our analyses agreed with the obligatory bathing water analyses.

In the water samples with gulls' faeces, the concentration of Enterococci was low. This indicates that contamination of bathing water samples with faeces from gulls does not necessarily lead to an elevated concentration and non-compliance with the regulations.

The similarity between two samples is determined on the basis of the similarity between pairs of Enterococcal isolates from the two samples. The similarity between samples is calculated as the number of pairs with a similarity above 0.965 divided by the total number of possible pairs. For example, the similarity between the isolates from the bathing water sample and the sample from the drainage canal taken on 21 August is 0.367. This means that of 420 possible pairs, 36.7% (=154 pairs) have a similarity above 0.965. A similarity of 1 means that all isolates from two samples are of the same phenotype. Similarities of 0 means that none of the isolates from the two samples have a similarity above 0.965.

The similarities between the three bathing water samples and the sample from the drainage canal were between 0.35 and 0.37, which is relatively high. This indicates that the contamination of the bathing water originates from the drainage canal. However, the local farmers told us that the pumping station had been out of service during the summer season, and an inspection of the pumping station showed that the direction of the water flow was from the lake into the drainage canal. The consequence of this is that the most probable explanation of the high similarity is that the lake is the source of contamination in the drainage canal.

The similarity between the bathing water samples from 21 August and 10 September 2003 was high, indicating that the source of contamination is the same on both occasions.

The similarity between the samples from the other potential sources of contamination and the bathing water does not clearly point out the source of contamination. Among the analysed samples, the samples of cattle faeces have the highest similarity to the bathing water samples. Therefore, the cattle faeces are considered to be the most likely source of contamination. However, in water samples taken close to the cattle watering location, the concentration of Enterococci was low, and the their phenotype was completely different from the phenotypes found in the bathing water samples. Therefore, the conclusion is subject to uncertainty. It is suggested that a supplementary investigation, in which samples are collected throughout the bathing water season, is carried out to confirm or reject the results of this investigation.

The PhPlate Rapid method could identify relations between the bathing water location, the drainage canal and the cattle faeces. However, other less explainable results were also obtained such as high similarity between the sewage infiltration plant and the pig slurry.

Based on the pilot investigation, it can be concluded that the PhPlate Rapid method can show similarities and differences between the phenotypes of the Enterococci from the bathing water location and the potential sources of contamination. The method is thus a promising technique for microbial source tracking.

The results indicated that the method in some cases is subject to uncertainty. The method requires experienced laboratory technicians to obtain optimal results.