CICEET Progress Report for the period 9/01/08 Through 2/15/09
Project Title: Automated Imaging and Classification System for Harmful Algal Bloom Detection
Principal Investigator(s): Dr. Robert J. Olson,
Additional Investigator(s): Dr. Heidi Sosik, Dr. Lisa Campbell
Project Start Date: 9/01/06
Figures |
||
|
||
|
||
|
||
|
Objectives
Our objectives for this reporting period were to continue the deployment of the Imaging FlowCytobot at the U. Texas-Marine Sciences Institute (UTMSI) at Port Aransas (at the entry to the Mission-Aransas NERR site), and to finalize the automated classifier for the local phytoplankton populations.
The data has been of high quality for most of the deployment, and has revealed several dramatic changes in phytoplankton community composition. Short-lived but intense blooms of several different dinoflagellates, diatoms, or flagellates were observed in succession from October through March (including Ebria, Brachydinium, Ditylum, Hemiaulus, Pyramimonas; and Dinophysis, Figure 4a and Figure 4b). Although several Karenia-like cells were observed during the fall, no bloom of Karenia brevis has been recorded in Texas waters this year. The cells we observed appeared to be Karenia mikimotoi (Figure 5). Dr. Campbell has manually examined samples at intervals over the deployment to compile a training set of images for the common phytoplankton groups observed, and Dr. Sosik has implemented automated classification of the image set as a whole, to obtain quantitative information about the different groups.
Observations from the deployment of IFCB on the Texas coast provided early warning for a Dinophysis bloom (Figure 1) that resulted in the first closure and recall of oyster harvests due to DSP above action level in the US. The alert occurred shortly before the Rockport Oysterfest, an annual event in the Port Aransas region that attracts up to 30,000 people, and so may have averted serious consequences to human health.
In early February 2008, Dinophysis cells (1-5 · mL-1) were detected by manual inspection of images; by late February, abundance estimates exceeded 200 cells · mL-1. Manual microscopy of water samples from the site confirmed that D. cf. ovum was the dominant species, with cell concentrations similar to those calculated from IFCB data, and toxin analyses showed that okadaic acid was present, which led to closing of shellfisheries. Analysis of the time series using automated image classification (extraction of image features and supervised machine learning algorithms) revealed a dynamic phytoplankton community composition. Before the Dinophysis bloom, Myrionecta rubra (a prey item of Dinophysis) was observed, and another dinoflagellate, Prorocentrum, was observed after the bloom. Dinophysis cell division rates, as estimated from the frequency of dividing cells, were highest at the beginning of the bloom and negligible after the peak of the bloom. Considered on a daily basis, cell concentration increased roughly exponentially up to the bloom peak, but closer inspection revealed that the increases generally occurred when the direction of water flow was into the estuary, suggesting the source of the bloom was offshore.
The results and data gathered during this reporting period have not changed the project objectives of the original proposal.
Dissemination activities during this reporting period (please include the number of participants where applicable).
Olson gave an invited presentation on "Plankton analysis by automated submersible flow cytometry" at MBARI (Moss Landing, CA), 19 Nov 2008.
Campbell presented "Automated continuous analysis of phytoplankton community composition by Imaging FlowCytobot reveals bloom dynamics anhd provides early warning of a HAB event", L. Campbell, R. J. Olson, H. M. Sosik at the ASLO meeting in Nice, Italy, January 2009.
Publications:
Campbell, L., R. J. Olson and H.M. Sosik. 2008. First toxic Dinophysis bloom observed in the Gulf of Mexico. Harmful Algae News 36: 10-11.
Campbell, L., R. J. Olson, H.M. Sosik, A. Abraham, D. Henrichs, E. Buskey, and C. Hyatt. First harmful Dinophysis (Dinophyceae, Dinophysiales) bloom in the US is revealed by automated imaging flow cytometry, to be submitted to J. Phycology.
WHOI's Intellectual Property Office has started discussions with BD Biosystems, Inc. (new owners of Cytopeia, Inc., with whom we are partners in a NOPP-funded project) about commercializing Imaging FlowCytobot
Project Objectives for Next Reporting Period
The project objectives for the next reporting period are to complete the automated classification of phytoplankton groups, and to write up the results for publication.
Our work plan to meet these objectives is the following:
- Campbell and Olson will continue to monitor the instrument's performance and perform maintenance as required.
- Sosik will complete modifications to the automated classification system for the species observed in Texas coastal waters.
- Analysis and dissemination of results will be carried out by all three PI’s.
No changes are necessary to the overall project timeline.
Expenditures
Expenditures to date are in the range anticipated.
End User Advisor Feedback
This report has been sent to our Advisor/End User for comment and their feedback will be incorporated when available.