2.3 Data Analysis
The 20-year trends in water quality variables were analyzed to allow for
(1) comparison against different environmental variables to understand
possible drivers of change and (2) evaluation of spatial variability in
water quality trends. Mann-Kendall (MK) testing and Sen Slope Estimators
were calculated and applied to annual average FC concentrations and
annual average salinity values for the water quality sampling stations,
and total annual precipitation for the terrestrial weather stations.
Fecal indicator bacteria concentrations are highly variable in space and
time, with samples collected consecutively over the span of a few
minutes sometimes varying by an order of magnitude over a few minutes
(Boehm, 2007). Accordingly, the annual arithmetic mean was selected to
represent the central tendency in long-term FC data, with the annual
time step selected to average out the effects of seasonal variation.
Similarly, using total annual precipitation informed us of the overall
freshwater load over the years and allowed us to explore long-term
drivers of baseflow FC concentrations. MK testing and Sen Slope
Estimation analysis have been used in numerous environmental,
hydrological, and water quality studies (Hirsch et al., 1982; Cailas et
al., 1986; Hipel et al., 1988; Zetterqvist, 1991; Burn et al., 2002;
Meals et al., 2011; Mustapha, 2013) due to their robustness against
non-normal data with missing values. Statistical analyses and mapping
were conducted in R with the ‘trend’ package version 1.1.4 (Pohlert,
2020) and the ‘sf’ package version 0.9 (Pebesma et al., 2022).