Hotspots and main drivers of fecal pollution in Neusiedler See, a large shallow lake in Central Europe.

To minimize the risk of negative consequences for public health from fecal pollution in lakes, the continuous surveillance of microbiological water quality parameters, alongside other environmental variables, is necessary at defined bathing sites. Such routine surveillance may prove insufficient to elucidate the main drivers of fecal pollution in a complex lake/watershed ecosystem, and it may be that more comprehensive monitoring activities are required. In this study, the aims were to identify the hotspots and main driving factors of fecal pollution in a large shallow Central European lake, the Neusiedler See, and to determine to what degree its current monitoring network can be considered representative spatially. A stochastic and geostatistical analysis of a huge data set of water quality data (~ 164,000 data points, representing a 22-year time-series) of standard fecal indicator bacteria (SFIB), water quality and meteorological variables sampled at 26 sampling sites was conducted. It revealed that the hotspots of fecal pollution are exclusively related to sites with elevated anthropogenic activity. Background pollution from wildlife or diffuse agricultural run-off at more remote sites was comparatively low. The analysis also showed that variability in the incidence of SFIB was driven mainly by meteorological phenomena, above all, temperature, number of sunny hours, and wind (direction and speed). Due to antagonistic effects and temporal undersampling, the influence of precipitation on SFIB variance could not be clearly determined. Geostatistical analysis did reveal that the current spatial sampling density is insufficient to cover SFIB variance over the whole lake, and that the sites are therefore in the most part representative of local phenomena. Suggestions for the future monitoring and managing of fecal pollution are offered. The applied statistical approach may also serve as a model for the study of other such areas, and in general indicate a method for dealing with similarly large and spatiotemporally heterogeneous datasets.