The distribution of submissions and positive tests over day of the week, month and year was examined using time series and seasonal trend loess (STL) decomposition (Hafen et al. 2009). With the exception of days of the week, all time series analyses were based on data aggregated to quarter section and month. Differences between submission rates per day of the week were examined with Pearson's chi-square test using R (version 2.14.0, R Development Core Team 2011). To examine seasonal variation, a time series was created for days of the week, months and years over the study period (2004–2012). We created these time series by ordering the E. coli-positive, total coliform-positive and total number of tests performed by equally spaced time intervals in Microsoft Excel and graphing the results. For the days of the week and months time series, data for all years was aggregated into the appropriate day of the week or month of the year. The data were also divided into three regions (north, central and south) based on Alberta Health Services administrative areas, and a separate time series was created for each region. These regions were also chosen to represent the latitude and climate gradient running north to south. Edward's test for seasonality was performed using WINPEPI version 11.18 (Abramson 2011) and peak dates were determined for the entire province and for the three regions (Table 1). STL decomposition (Figures 1 and 2) was performed using R (version 2.14.0, R Core Development Team 2011) to decompose the time series into components: seasonal, trend and residuals, using a locally weighted non-parametric regression (Cleveland et al. 1990).