in a Changing Ocean
Monitoring of the Trichodesmium blooms in the Canary Islands
The purpose of this project is to monitor Trichodesmium blooms, consolidate the predictive model, control the beneficial and/or harmful effects of blooms and control pathogens associated to Trichodesmium mucilages washed ashore.
This objective falls within the priority areas defined in the Cooperation Framework Agreement signed on April 26, 2019 between the University of Las Palmas de Gran Canaria and the Department of Territorial Policy, Sustainability and Security (currently called the Department of Ecological Transition, Fight against Climate Change and Territorial Planning), specifically in Stipulation 2, section a) “Joint studies and research” and section g) “Promote agreements for the installation of long-term programs duration in the Canary Islands ”.
Describe the spatiotemporal dynamics of Trichodesmium in the waters around the Canary Islands: origin, maintenance and disappearance of blooms and their relationship with environmental variables.
To accomplish this objective, on-site monitoring of Trichodesmium populations will be carried out throughout the year (at least once a month).
Study the fertilization effect of Trichodesmium blooms on the marine ecosystem.
This study will be carried out by combining the following methods: field samplings and controlled experiments:
Field samplings will be carried out at sea at high spatial resolution (from 10 to 100 m), collecting Trichodesmium samples from bloom formations, to observe the fertilization effect by nutrients and organic matter associated with these blooms
Controlled experiments will consist on adding extracts of supernatant from cell exudates that they release when dying, from natural samples from the coast.
Development and validation of the predictive model on the dynamics of formation and collapse of these blooms.
The prediction of algal episodes includes different predictive model architecture methods. To this date, the architecture of a predictive model based on neural networks (“Machine Learning Modeling Techniques”) has been developed requiring long time series of data (at least 10-20 years) in order to be effective, since the model "learns" from the in situ results.
For this reason, we propose a different model approach, using short time series (since 2017) and multidimensional analysis, which quantify Trichodesmium blooms probability of occurrence as a function of certain representative variables (key factors for bloom occurrence).