In order to detect if differences
were consistent across river sections' and seasons in each year's categories we then performed pair-wise tests for termsreservoirsandseason
(nested in year). After this, we divided the matrix in three new matrices,
one comprising sampling sites located upstream the reservoirs
(upstream, 84 samples), another comprising the sampling site located
intermediate the reservoirs (intermediate, 14 samples), and a last
one comprising sampling sites located downstream the reservoirs
(downstream, 68 samples).