Figure 2: Comparison of the three sampling positions. Left: Venn diagram for the different combinations of sampling positions. For each combination read proportion in relation to overall macroinvertebrate reads is given in percent, numbers of species shown in brackets. Right: species richness over time for each sampling position. Points represent pure species richness, solid lines represent the fitted values of the GAM.
Additionally, the NMDS analysis showed no differences in species or OTU beta-diversity between sampling positions based on ANOSIM tests (p-value: 0.7371/0.7079) (Figure S5 & S6). Based on these results, samples from the different sampling positions were merged for each sampling day and subsequent analyses focused on sampling time and seasonal differences. NMDS analysis for the merged samples revealed a strong impact of sampling season on community structure (Figure 3 & S7). Samples from following years cluster closely together e.g. 07.06.2017/06.06.2018 and 23.05.18/24.05.17. Summer and winter samples were separated on the first axis, spring and autumn samples on the second axis, producing a cyclic temporal beta-diversity pattern resembling a ‘seasonal clock’. When temporal beta-diversity was partitioned into turnover and nestedness, turnover was the driving factor for change in beta-diversity while nestedness was of less importance (Figure S8). Over the year, turnover increased over the winter samples and decreased again in summer with nestedness showing the opposite trend, reaching its highest value in summer 2018 (Figure S8).