Automated imaging in Long-Term Ecological Research reveals taxon-specific bloom anomalies in phytoplankton

Coastal pelagic waters support complex and dynamic ecosystems that depend on planktonic food webs. These systems are challenging to observe at the appropriate spatial, temporal, and taxonomic scales. We are working to meet this challenge with automated microscopic imaging at the Northeast U.S. Shelf Long-Term Ecological Research (NES-LTER) site. For over a decade, we have deployed Imaging FlowCytobot both at a nearshore cabled observatory and on-board research vessels surveying across the shelf seasonally. With this vast dataset (>1.5 billion images) and machine-learning for image classification, we are now able to document where and when unusual blooms of specific taxa or groups of phytoplankton occur. These atypical blooms are surprisingly common in the dataset, but also highly variable both in which taxa occur and in spatial and temporal location and extent. This presentation will highlight examples of decade-scale outlier abundances of specific diatoms, dinoflagellates, and haptophytes. We will show that some events appear to be nearshore phenomena with timescales of weeks while others are localized in mid- or outer- shelf waters over similar timeframes, and furthermore that certain massive, unprecedented blooms can persist for months and extend over 100s of kilometers along the NES. While we have an evolving understanding about factors promoting some of these types of blooms, in many cases they remain mysterious to explain.

Presentation Preference: Either

Primary Presenter: Heidi Sosik, Woods Hole Oceanographic Institution (hsosik@whoi.edu)

Authors:

Heidi Sosik, Woods Hole Oceanographic Institution  (hsosik@whoi.edu)

Emily Peacock, Woods Hole Oceanographic Institution  (epeacock@whoi.edu)

E Taylor Crockford, Woods Hole Oceanographic Institution  (ecrockford@whoi.edu)

Miraflor Santos, Woods Hole Oceanographic Institution  (msantos@whoi.edu)