The Ecological Dynamics of Fecal Contamination and Salmonella Typhi and Salmonella Paratyphi A in Municipal Kathmandu Drinking Water

Abstract

One of the UN sustainable development goals is to achieve universal access to safe and affordable drinking water by 2030. It is locations like Kathmandu, Nepal, a densely populated city in South Asia with endemic typhoid fever, where this goal is most pertinent. Aiming to understand the public health implications of water quality in Kathmandu we subjected weekly water samples from 10 sources for one year to a range of chemical and bacteriological analyses. We additionally aimed to detect the etiological agents of typhoid fever and longitudinally assess microbial diversity by 16S rRNA gene surveying. We found that the majority of water sources exhibited chemical and bacterial contamination exceeding WHO guidelines. Further analysis of the chemical and bacterial data indicated site-specific pollution, symptomatic of highly localized fecal contamination. Rainfall was found to be a key driver of this fecal contamination, correlating with nitrates and evidence of S. Typhi and S. Paratyphi A, for which DNA was detectable in 333 (77%) and 303 (70%) of 432 water samples, respectively. 16S rRNA gene surveying outlined a spectrum of fecal bacteria in the contaminated water, forming complex communities again displaying location-specific temporal signatures. Our data signify that the municipal water in Kathmandu is a predominant vehicle for the transmission of S. Typhi and S. Paratyphi A. This study represents the first extensive spatiotemporal investigation of water pollution in an endemic typhoid fever setting and implicates highly localized human waste as the major contributor to poor water quality in the Kathmandu Valley.

Fig 1. The physical and chemical properties of water from ten sampling locations in Kathmandu, Nepal.

a) Photograph of child collecting water from a traditional stone waterspout (dhunge dhara) in Kathmandu. b) A map showing the location of the ten sampling locations in Kathmandu. The locations are color-coded corresponding to other figures and the site of Patan hospital (the healthcare facility for those with typhoid fever) is highlighted by the H symbol. (See Table 1 and Baker et al. [6] for more details regarding the sampling locations and the mapped area). c) Scaled principal components analysis of the physical and chemical properties of water samples from the ten sampled locations, showing the first two principal components (PC1-2). Individual water sample are identified by dots colored according to their sampling location (see Fig 1a). Inertia ellipses indicate the overall distribution of each water source. Labeled arrows indicate the relative contributions of the variables to the principal components, with longer arrows reflecting larger contributions. The screeplot of eigenvalues (inset) indicates the amount of variation contained in the different principal components, with PC1 and PC2 indicated in black.

Fig 2. The quantification of coliforms and DNA from Salmonella serovars Typhi and Paratyphi A in water samples from ten locations in Kathmandu, Nepal.

a) Boxplots of the MPN counts (log10 CFU/ml) aggregated by water sources type (SS, stone spout; SW, sunken well; PS, piped supply) and within each individual location (numbered on x-axis). Boxes and horizontal lines represent the interquartile ranges and medians, respectively and whiskers represent 90% of data range. The median MPN measurements in water from stone spouts; sunken wells and piped supplies were significantly different; as determined by Kruskal Wallis test (p<0.001; shown by the asterisk). b) Scatterplots showing the number of copies of S. Typhi (red circles) and S. Paratyphi A (blue circles) gene targets calculated to be in each water sample (by realtime PCR amplification and quantification using a standard curve) against the week of sampling. The red and blue shaded lines show the best fit through time of gene copy quantification trends for S. Typhi and S. Paratyphi A, respectively. c) Principal components analysis of the weekly presence/absence profiles of S. Typhi and S. Paratyphi A gene targets in water samples, shaded by weekly rainfall (see key). Time is represented on the x-axis. The first principal component, representing fluctuations in S. Typhi and S. Paratyphi A DNA relative abundance, is plotted on the y-axis.

Fig 3. 16S rRNA gene surveying of gastrointestinal bacterial taxa found in stone spout and well water samples in Kathmandu, Nepal.

a) Screeplot showing the eigenvalues of the PCA of 16S rRNA gene data, with retained axes in blue corresponding to 80% of the variation in 519 identified OTUs from fecal bacterial families (Bacteroidaceae, Clostridiaceae, Enterobacteriaceae, Erysipelotrichaceae, Lachnospiraceae, Lactobacillaceae, Prevotellaceae, Ruminococcaceae and Veillonellaceae) identified in 93 water samples from location 2 (stone spout) and location 5 (sunken well). The main graph only represents the first 30 eigenvalues (full graph provided in inset). b) Diversity represented by the OTUs with large contributions to the retained PCA axes. PCA axes are represented on the x-axis, with a width proportional to the corresponding diversity (eigenvalue). The y-axis represents the amount of diversity retained by retaining only OTUs with contributions of at least 1% (6 taxa), 5% (5 taxa), 50% (4 taxa) or 75% (2 taxa), indicated by differing shades of blue. The total surface of a given color is proportional to the fraction of the total diversity represented by this set of taxa. The red dashed line identifies the set of 6 retained taxa plotted in Fig 3c, representing 76% of the total variation in the entire data. c) OTU composition of the water samples, showing the relative frequencies of the six most structuring fecal bacterial taxa identified in Fig 3b (Enterobacteriaceae (OTUs 00024, 00185 and 01479), Bacteroidaceae (OTU00979), Clostridiaceae (OTU00263), and Prevotellaceae (OTU 2149)). Their relative abundance (y-axis) is represented through time in location 2 (stone spout) and location 5 (sunken well), the two locations with the greatest estimated coliform contamination by MPN. Empty bars correspond to missing or failed samples. d) A Discriminant Analysis of Principal Components (DAPC) of the 16S rRNA gene identifying combinations of the 519 gastrointestinal OTUs differing the most between the stone spout (blue) and the sunken well (red). The OTUs exhibiting the greatest variation between these locations were OTU00263; Clostridium, OTU00185; Enterobacteriaceae, OTU00979; Bacteroides and OTU00024, Enterobacteriaceae).