Study of the impact of TIN/PO4 ratio on mucilage formation in the northern Adriatic using regression trees

Authors

  • Goran VOLF
  • Nataša ATANASOVA
  • Boris KOMPARE
  • Robert PRECALI
  • Nevenka OŽANIĆ

Keywords:

TIN/PO4 ratio, mucilage events, machine learning, regression trees, northern Adriatic

Abstract

     The north-western part of the northern Adriatic (NA) exhibits eutrophic to mesotrophic characteristics with recurrent algal blooms and quite unpredictable mucilage events. To contribute to the understanding of the mucilage events in the NA a machine learning algorithm for induction of regression trees was applied on a long-term data-set comprising physical, chemical and biological parameters, measured at six stations on the profile from the Po River delta (Italy) to Rovinj (Croatia). A model describing the connection between the TIN/PO4 ratio, considered as a necessary factor and sometimes even a trigger for mucilage events, and the environmental conditions in NA, was elaborated. The model for TIN/PO4 ratio confirmed the assumption that the mucilage events are connected with this ratio, e.g. mucilage events coincides with its high values. This finding indicates that at certain levels of phosphorus limitation (from TIN/PO4 ratio) mucilage event frequency increases. The model also reveals that salinity and temperature are responsible for the changes of the TIN/PO4 ratio and gives an insight on their threshold values which lead to high values of this ratio, further related to mucilage events.

Downloads

Published

15.12.2015

Issue

Section

Original article