Hypothesis 4: Lower oxygen concentration in summer will have a negative effect on the functional richness.
Lower oxygen content in summer was expected to reduce the functional diversity. H4 was expected to be true in open gulf; therefore, only open gulf data were used.

Sensitivity tests

To assure that none of the long-term changes were caused by sampling or by taxonomic artefacts, sensitivity analyses were carried out, in which the extent of the repositioning of communities at the PCoA space was assessed while omitting one taxon at the time from the data. This analysis was done by same subsets of data as shown in Figure 3. Diagnostics plots were drawn for the most influential taxa (supplementary Fig. X) that were selected after omitting a particular taxon. The A1 or A2 values of the selected samples had a Pearson’s correlation with full community A1 or A2 values <0.9). No artefacts (i.e., sudden unexplainable appearance or disappearance of the taxon) were revealed during this test, although several taxa had a strong impact on the long term trend. Altogether 11 influential taxa were identified during this test in 19 cases, and in 5 cases, a long-term trend became notably stronger after excluding a single taxon (e.g. Synchaeta sp. and Acartia sp. in open sea area in May along PCoA axis 1 (Fig. S1c), or Bosmina sp. in open sea in August, along PCoA axis 1, Fig. S1g). All analyses were performed using the R software (version 3.4.0) and RStudio (version 1.1.442; R Development Core Team 2012).

Results