Abstract:
This study tested and evaluated the use of three Geographic information Systems (GIS)-based computer models for their suitability in spatially predicting concentrations of Campylobacter in the Ruamahanga River catchment, a primarily agricultural catchment in the lower North Island of New Zealand. Two of these models evaluated the potential for delivery of Campylobacter bacteria to streams while the third evaluated the transportation and fate of bacteria within streams. This modelling was supplemented with two field sample sets to determine how well the predictions of the models matched the observed field values.
Campylobacter was found in the Ruamahanga catchment, with 67% of all samples in the second sample set being positive for Campylobacter, similar to results obtained in other New Zealand-based studies. The concentrations of Campylobacter in this study were frequently too low to pose a risk to recreational users at swimming sites; however, this study and other studies show elevated concentrations of Campylobacter in some instances. These findings show that there is still a need to identify the conditions that cause increases in the concentration of Campylobacter at swimming sites. A strong negative relationship (x2 = 14.04, df = 2, p = 0.0008) was found between elevated and unacceptable concentrations of Campylobacter and clear sky conditions, underscoring the sensitivity of Campylobacter to UV light. A weaker positive correlation ( x2 = 6.71, df = 2, p = 0.035 ) was found between elevated concentrations of Campylobacter and the presence of livestock. Weaker correlations were also found with BOD and with nitrate concentrations, which are common indicators of water quality.
Neither of the delivery models showed highly significant correlations of measured Campylobacter concentrations with any of their predictions, making them unsuitable for day-to-day predictions of Campylobacter concentrations. This poor correlation, coupled with an absence of precipitation during the sampling period suggested another method of introduction of Campylobacter into waterways during the sampling rather than runoff. It is likely that much of the bacteria entered waterways though either subsurface flow or direct deposition by cattle that had access to a stream. One of the delivery models and the in-stream model greatly under-predicted the concentrations of Campylobacter in streams suggesting that either Campylobacter is entering streams at much higher concentrations than predicted or that the amount of bacteria being shed by livestock is greatly underestimated for New Zealand conditions. Further study of important parameters for the in-stream model will allow further testing and validation.
