Does land use affect pathogen presence in New Zealand drinking water supplies?. (15th October 2020)

• Record Type: Journal Article
• Authors: Phiri, Bernard J.; Pita, Anthony B.; Hayman, David T.S.; Biggs, Patrick.J.; Davis, Meredith T.; Fayaz, Ahmed; Canning, Adam D.; French, Nigel P.; Death, Russell G.
• Is Part Of: Water research. Volume 185(2020)
• Publication Date: 2020-10-15
• Subjects: Water -- Pollution -- Research -- Periodicals; 363.7394
• DOI: 10.1016/j.watres.2020.116229
• Languages: English
• ISSNs: 0043-1354

Abstract

Highlights: Pathogens were monitored in 16 New Zealand waterways supplying drinking water. Cryptosporidium and Giardia were prevalent at sites with high numbers of ruminants. Campylobacter and E. coli were prevalent in pasture catchments and after rain. Catchment management is a vital tool for limiting outbreaks of waterborne disease. Abstract: Four microbes ( Campylobacter spp., Escherichia coli, Cryptosporidium spp. and Giardia spp.) were monitored in 16 waterways that supply public drinking water for 13 New Zealand towns and cities. Over 500 samples were collected from the abstraction point at each study site every three months between 2009 and 2019. The waterways represent a range from small to large, free flowing to reservoir impoundments, draining catchments of entirely native vegetation to those dominated by pastoral agriculture. We used machine learning algorithms to explore the relative contribution of land use, catchment geology, vegetation, topography, and water quality characteristics of the catchment to determining the abundance and/or presence of each microbe. Sites on rivers draining predominantly agricultural catchments, the Waikato River, Oroua River and Waiorohi Stream had all four microbes present, often in high numbers, throughout the sampling interval. Other sites, such as the Hutt River and Big Huia Creek in Wellington which drain catchments of native vegetation, never had pathogenic microbes detected, or unsafe levels of E. coli . Boosted Regression Tree models could predict abundances and presence/absence of all four microbes with good precision using a wide range of potential environmental predictors covering land use, geology, vegetation, topography, and nutrient concentrations. Models were more accurate for protozoa than bacteria but did not differ markedly in their ability to predict abundance or presence/absence. Environmental drivers of microbe abundance or presence/absence also differed depending on whether the microbe was protozoan or bacterial. Protozoa were more prevalent in waterways with lower water quality, higher numbers of ruminants in the catchment, and in September and December. Bacteria were more abundant with higher rainfall, saturated soils, and catchments with greater than 35% of the land in agriculture. Although modern water treatment protocols will usually remove many pathogens from drinking water, several recent outbreaks of waterborne disease due to treatment failures, have highlighted the need to manage water supplies on multiple fronts. This research has identified potential catchment level variables, and thresholds, that could be better managed to reduce the potential for pathogens to enter drinking water supplies.