Degree Granting Department
Estuaries, Florida, Ocean Optics, Spectroscopy, Tampa Bay, Taxonomy
Phytoplankton serve as the bottom of the marine food web and therefore play an essential role in marine ecosystems. On the other hand, coastal phytoplankton communities can adversely affect the marine ecosystem and humans. A variety of techniques have been developed to measure and study phytoplankton, including in situ methods (e.g., flow cytometry) and laboratory methods (e.g., microscopic taxonomy). These provide accurate measurements of phytoplankton taxa and concentrations, yet they are limited in space and time, and synoptic information is difficult to obtain with these techniques.
Optical remote sensing may provide complementary information for its synoptic nature, as demonstrated by satellite estimates of major phytoplankton taxa in major ocean basins. It has remained a challenge, however, for coastal and estuarine waters due to their optical complexity. One pioneering work relied on hyperspectral absorption spectra of phytoplankton pigments (Millie et al., 1995), from which Gymnodinium breve (i.e., Karenia brevis) blooms on the West Florida shelf could be detected and quantified in situ. However, whether a similar approach can be developed for estuarine waters where toxic blooms are often found is still unknown. Thus, the objective of this study is to test and develop an approach to classify major phytoplankton taxa found in two estuaries in Florida, U.S.A., based on optical analysis of the phytoplankton absorption spectra.
In this study, over 250 surface water samples were collected on numerous cruise surveys from two Florida estuaries (Tampa Bay, ∼1000 km2 on the west coast; and the Indian River Lagoon, ∼900 km2 on the east coast). The samples were filtered and then processed using standard NASA protocols to determine 1) their spectral absorption coefficients due to phytoplankton pigments, aph (λ) (m-1), and 2) their chlorophyll a concentrations (mg m-3). aph (λ) was further normalized by Chl a, resulting in chlorophyll-specific absorption coefficient, a aph∗ (λ) (m2 mg-1). For each sample, phytoplankton cell counts were enumerated by the Florida Wildlife Conservation Commission (FWC) Fish and Wildlife Research Institute (FWRI) through microscopic taxonomy. The a aph∗ (λ) data were then categorized based on the dominant phytoplankton taxa, and were separated as either bloom or non-bloom using a 100,000 cell∕L threshold of the dominant taxa. Three techniques were tested for classifying phytoplankton taxa using absorption spectra; a first derivative summation, a relative height analysis, and an integration analysis. The integration technique proved to be the most successful of the three. This technique performed an integration of a aph∗ (572-600nm) against a linear baseline, and yielded an 81% success rate (13 of 16 samples) and 9% false positive rate (13 of 144 samples) in separating blooms of the dinoflagellate Pyrodinium bahamense from other bloom and non-bloom taxa found in the Tampa Bay estuary. The same integration technique, but with the wavelength range shifted to 471 nm - 490 nm, was also applied to the samples collected in the Indian River Lagoon estuary from summer 2011 to study the green flagellate of the class Pedinophyceae.. The results showed an 80% success rate (8 of 10 samples) and a 0.5% false positive rate (1 of 156 samples) in separating the Pedinophyceae bloom taxa from other bloom and non-bloom taxa found in both the Indian River Lagoon and Tampa Bay.
The number of bloom samples was relatively low (16 from Tampa Bay and 10 from IRL). Thus, the results from this study are preliminary and will require more sampling in order to further develop this technique to a practical method for field use. However, the results obtained from this study are comparable to those from other techniques for classification of phytoplankton taxa, for example, BreveBuster, SIPPER, FlowCAM, and satellite ocean color remote sensing of the open ocean. Yet this technique extends to optically complex estuarine waters, and therefore may represent a step towards the ultimate goal of applying satellite remote sensing in characterizing phytoplankton taxa in estuaries. Once confirmed with more samples from the same two estuaries as well as from other estuaries, an immediate next step may be the implementation of in situoptical instruments on either buoys (e.g., MARVIN in Tampa Bay) or flow-through systems to provide continuous characterization of major phytoplankton taxa in the two estuaries.
Scholar Commons Citation
Sensi, Daniel Tyler, "Optical Detection and Classification of Phytoplankton Taxa through Spectral Analysis" (2012). Graduate Theses and Dissertations.