Phytoplankton Time-Series in a LTER Site of the Adriatic Sea: Methodological Approach to Decipher Community Structure and Indicative Taxa

Abstract

In the shallow and landlocked northeast Adriatic Sea, environmental factors have changed in recent decades. Their influence on seasonal and inter-annual variability of phytoplankton has been documented in the recent literature. Here, we decipher the long-term variability of phytoplankton phenology at a Long-Term Ecological Research site (Gulf of Trieste, Slovenia). Structural changes in the phytoplankton community (period 2005–2017) were analysed using a multivariate protocol based on Bayesian clustering. The protocol was modified from the literature to fit the needs of the study, using correspondence analysis and k-means clustering. A novel index for ordination and selection of taxa based on frequency and evenness was developed. The Total Inertia analysis showed that this index better preserved the available information. Typical seasonal assemblages were highlighted by applying the Indicative Value index in conjunction with likelihood ratio values. We obtained a rough picture of the seasonal separation of the diatom-dominated community from the mixed community and a refined picture of the phenology of the assemblages and bloom events. The spring diatom peak proved to be inconstant and short-lived, while the autumn bloom was generally long and diverse. As expected for nearshore environments, the average life span of the assemblages was found to be short-periodic (2–4 months). The second part of the year and the last part of the series were more prone to changes in terms of typical assemblages.