Since the first reported occurrence of Legionnaire disease in 1976 , infectious diseases caused by Legionella pneumophila have been successively reported worldwide [2-8]. Currently, Legionella species are considered common causes of community-acquired and nosocomial pneumonia [9-11]. Water environments appear to be the natural habitat and to serve as amplifiers for Legionella species, which are facultative intracellular parasites [12, 13]. Roof-harvested rainwater has attracted significant attention as a potential alternative source of potable and nonpotable water, especially in regions where water is scarce . To encourage the use of roof-harvested rainwater for domestic purposes, governmental bodies of many countries, such as Australia [15, 16], Denmark , and New Zealand , have been providing subsidies to residents. However, there are insufficient data concerning the microbiological quality of roof-harvested rainwater and its potential health risks.
In Japan, there has also been increasing recognition in both urban and rural areas that it is important to utilize rainwater with the aim of effective water reuse and water conservation. Accordingly, many administrative agencies in Japan have also been encouraging installation of rainwater tanks by subsidizing the costs of installation. As a result, many more rainwater tanks have been installed in Azumino city, Nagano prefecture, a rural area of Japan.
The increasing installation of rainwater tanks in urban and/or rural environments necessitates greater understanding of the quality of water they are able to supply. Although there are few published studies on the physico-chemical properties of harvested rainwater [19-23], various factors including topography, climate, atmospheric temperature and surrounding circumstances are all considered to influence the quality of rainwater stored in such tanks.
To obtain current information on the microbiology of tank rainwater in rural area of Japan, we investigated water samples from rainwater tanks for microbiological contamination, focusing on the presence of Legionella species. We simultaneously assessed the pH, COD, atmospheric temperature, climate, number of heterotrophic bacteria and topography, in order to evaluate their mutual relationships.
The objectives of this study were to assess the presence of Legionella and to compare it with various variables concerning harvested rainwater in samples taken during spring, summer and autumn.
- Top of page
- MATERIALS AND METHODS
Legionella species are ubiquitous in freshwater environments and have been isolated from various aquatic habitats, including man-made water distribution systems [10-13]. Legionella species have frequently been isolated in association with amoebae, which may serve as a host for this species. The primary route of infection with Legionella species is considered to be inhalation. Roof-harvested rainwater in household tanks can be used in many aspects of daily life, such as washing family cars, sprinkling gardens with water and watering flowers and plants, all of which may result in the formation of potentially infectious aerosols.
Among the water samples from 43 rainwater tanks of 40 users, we found that only one sample was inhabited by 80 CFU/100 mL of Legionella. Therefore, almost all rainwater samples that were PCR positive for Legionella were below the detection limit for Legionella cells, that is, less than 10 CFU/100 mL; it should be noted that PCR assays detect both living and dead Legionella.
It should be noted that we found that approximately one third of storage tanks were contaminated with Legionella. In detail, among the 40 households investigated in this survey, we demonstrated five to be PCR-positive for Legionella on all three test occasions, seven were positive twice and 11 were PCR-positive for Legionella only once. That is, we found that Legionella species inhabited 23 of 40 household rainwater tanks in Azumino city. As shown in Figure 1, an interesting finding was that samples PCR-positive for Legionella were from households situated in districts A and B, which are both concentrated on the conventional railway Oito line and along national highway Route 147. On the other hand, most households in districts C and D are located in submontane districts and were PCR-negative for Legionella. However, we observed no distinctive differences in heterotrophic bacterial counts and COD values between districts A and B and districts C and D. As demonstrated in Table 2 in conjunction with Figure 2, the PCR-positive percentages for Legionella varied according to the amount of precipitation. Our results suggest that Legionella contamination is related to the amount of precipitation and the location of the rainwater tanks. Indeed, in high-traffic districts, rainwater tanks are apt to be exposed to the spreading of Legionella-contaminated soil or rainwater . Water in transient puddles formed on roads on rainy days might be splashed into the air by moving vehicles on national highway Route 147. During ongoing sunny days, hetrotrophic bacteria multiply profusely. On the other hand, with ongoing rain, heterotrophic bacterial counts decreased, probably because of dilution by continual flow of rainwater. This would lead to degradation of rainwater in tanks allowing habitation by the amoebae required for proliferation of Legionella.
Although not presented in the Results section, pH of the rainwater was distributed unevenly between 3.6 and 7.0, but was approximately constant over the three measurements in each household; we observed no obvious correlations between heterotrophic cell counts and PCR-positivity for Legionella. Variations in pH of rainwater in household storage tanks probably depend on the amount of automobile emissions and the quality of roofs, walls and other building materials.
Legionella species are known to infect the trophozoite form of free-living amoebae and to replicate intracellularly at temperatures over 25°C . In contrast, at temperatures below 20°C, Legionella species are actively digested by amoebae and eliminated from them. The temperatures during the investigation periods of June and August were mostly over 20°C; we measured comparatively higher temperatures in August. Therefore, at least during this period, we consider that atmospheric temperatures probably did not influence detection of Legionella.
Identification of indicator microorganisms or other measureable indicators of Legionella pollution is desirable. We attempted to examine the relationships between our findings and the presence of Legionella. We found that both the number of heterotrophic bacteria in rainwater tanks and COD values appeared to correlate with the presence of Legionella species during the period between June and October.
When we converted the numbers of heterotrophic bacterial cells with PCR-positivity for Legionella to logarithmic scales, almost all heterotrophic bacterial cell counts in rainwater samples that were PCR-positive for Legionella were over 104 CFU/mL, as shown in Figure 3. We proved this finding was significant by means of null hypothesis using non-parametric Wilcoxon's signed-rank sum test with a significance level of 5%. In addition, we demonstrated clean linearity by means of Q–Q plot test on log-normal probability paper, as shown in Figure 4. The log-normal distribution was moreover verified both by means of Kolmogorov–Smirnov and Shapiro–Wilk tests. That is, they were confirmed according to the summary statistics as a normal distribution with an average of 5.2581, normal distribution of standard deviation of 1.0572 from unbiased variance. In this analysis, the position of 4 (104 CFU/mL of heterotrophic cell counts) was standardized by the following formula: z = (4−5.2581)/1.0572 = −1.194. Thus, the probability of distribution of less than 104 CFU/mL became 0.117 of the original value. In other words, 88.3% (1−0.117 = 0.883) of the samples that were PCR-positive for Legionella was distributed in the rainwater samples with heterotrophic bacterial cell counts of more than 104 CFU/mL.
These results were fairly consistent with a recent survey that used multivariate regression to show a relationship between the presence of Legionalla and heterotrophic cell counts of over 1 × 105 CFU/100 mL (i.e., 103 CFU/mL) .
Moreover, that the positive percentages for detection of Legionella species were demonstrated to be reduced when COD was more than 5 mg KMnO4/L, as shown in Table 3, is of considerable interest. This finding was statistically investigated by means of Pearson's product-moment correlation coefficient test and found to be significant with a coefficient of correlation value of −0.9970.
As far as we know, there have been few reports concerning index organisms associated with contamination with Legionella cells. Adequate numbers of amoebae would be required to achieve substantial multiplication of intracellular parasitic Legionella. In addition, adequate cell counts of heterotrophic bacteria are needed for substantial proliferation of amoebae. However, the higher the heterotrophic bacterial count, the higher the COD value. An increase in COD values implies deterioration of water conditions; this could be unfavorable for inhabitation by amoebae. Amoebae might have difficulty proliferating in water conditions with COD values of >5 mg KMnO4/L, as shown in Table 3.
Therefore, we consider that a threshold number (less than 104 CFU/mL) of heterotrophic bacteria as well as an amount of COD (more than 5 mg KMnO4/100 mL) might be useful environmental management indicators of the presence of Legionella cells in rainwater tanks during the period between June and October, at least in Azumino, Japan. Much more accumulation of such data is needed to verify the accuracy of our findings.