Application of a convolutional neural network to improve automated early warning of harmful algal blooms.

Continuous monitoring and early warning together represent an important mitigation strategy for harmful algal blooms (HAB). The coast of Texas experiences periodic blooms of three HAB dinoflagellates: Karenia brevis, Dinophysis ovum, and Prorocentrum texanum. A plankton image data set acquired by an Imaging FlowCytobot over a decade of operation was used to train and evaluate two new automated image classifiers. A 112 class, random forest classifier (RF_112) and a 112 class, convolutional neural network classifier (CNN_112) were developed and compared with an existing, 54 class, random forest classifier (RF_54) already in use as an early warning notification system. Both 112 class classifiers exhibited improved performance over the RF_54 classifier when tested on three different HAB species with the CNN_112 classifier producing fewer false positives and false negatives in most of the cases tested. For K. brevis and P. texanum, the current threshold of 2 cells.mL-1 was identified as the best threshold to minimize the number of false positives and false negatives. For D. ovum, a threshold of 1 cell.mL-1 was found to produce the best results with regard to the number of false positives/negatives. A lower threshold will result in earlier notification of an increase in cell concentration and will provide state health managers with increased lead time to prepare for an impending HAB.

Characterization of Dinophysis spp. (Dinophyceae, Dinophysiales) from the mid-Atlantic region of the United States1.

Due to the increasing prevalence of Dinophysis spp. and their toxins on every US coast in recent years, the need to identify and monitor for problematic Dinophysis populations has become apparent. Here, we present morphological analyses, using light and scanning electron microscopy, and rDNA sequence analysis, using a ~2-kb sequence of ribosomal ITS1, 5.8S, ITS2, and LSU DNA, of Dinophysis collected in mid-Atlantic estuarine and coastal waters from Virginia to New Jersey to better characterize local populations. In addition, we analyzed for diarrhetic shellfish poisoning (DSP) toxins in water and shellfish samples collected during blooms using liquid-chromatography tandem mass spectrometry and an in vitro protein phosphatase inhibition assay and compared this data to a toxin profile generated from a mid-Atlantic Dinophysis culture. Three distinct morphospecies were documented in mid-Atlantic surface waters: D. acuminata, D. norvegica, and a "small Dinophysis sp." that was morphologically distinct based on multivariate analysis of morphometric data but was genetically consistent with D. acuminata. While mid-Atlantic D. acuminata could not be distinguished from the other species in the D. acuminata-complex (D. ovum from the Gulf of Mexico and D. sacculus from the western Mediterranean Sea) using the molecular markers chosen, it could be distinguished based on morphometrics. Okadaic acid, dinophysistoxin 1, and pectenotoxin 2 were found in filtered water and shellfish samples during Dinophysis blooms in the mid-Atlantic region, as well as in a locally isolated D. acuminata culture. However, DSP toxins exceeded regulatory guidance concentrations only a few times during the study period and only in noncommercial shellfish samples.

Continuous automated imaging-in-flow cytometry for detection and early warning of Karenia brevis blooms in the Gulf of Mexico.

Monitoring programs for harmful algal blooms (HABs) typically rely on time-consuming manual methods for identification and enumeration of phytoplankton, which make it difficult to obtain results with sufficient temporal resolution for early warning. Continuous automated imaging-in-flow by the Imaging FlowCytobot (IFCB) deployed at Port Aransas, TX has provided early warnings of six HAB events. Here we describe the progress in automating this early warning system for blooms of Karenia brevis. In 2009, manual inspection of IFCB images in mid-August 2009 provided early warning for a Karenia bloom that developed in mid-September. Images from 2009 were used to develop an automated classifier that was employed in 2011. Successful implementation of automated file downloading, processing and image classification allowed results to be available within 4 h after collection and to be sent to state agency representatives by email for early warning of HABs. No human illness (neurotoxic shellfish poisoning) has resulted from these events. In contrast to the common assumption that Karenia blooms are near monospecific, post-bloom analysis of the time series revealed that Karenia cells comprised at most 60-75 % of the total microplankton.

Morphology and Phylogeny of Prorocentrum texanum sp. nov. (Dinophyceae): A New Toxic Dinoflagellate from the Gulf of Mexico Coastal Waters Exhibiting Two Distinct Morphologies.

A new planktonic species of Prorocentrum is described from the Gulf of Mexico. First observed with the Imaging FlowCytobot, Prorocentrum texanum sp. nov. was characterized using LM, SEM, and TEM along with sequencing of the SSU, LSU, and ITS ribosomal regions and the mitochondrial cob and cox1 regions. P. texanum sp. nov. is a round to oval bivalvate dinoflagellate, with a prominent anterior, serrated solid flange on periflagellar a platelet and an opposing short, flat flange on the h platelet. The periflagellar area consists of 10 platelets. Both left and right valves have shallow round depressions and two-sized valve pores. The anterior ejectosome pore pattern differs between the left and right valve in relation to the periflagellar area and margins. Ten to eleven rows of tangential ejectosome pores are present on each valve. P. texanum sp. nov. has two varieties which exhibit distinct morphotypes, one round to oval (var. texanum) and the other pointed (var. cuspidatum). P. texanum var. cuspidatum is morphologically similar to P. micans in surface markings, but is smaller, and has a serrated periflagellar flange, and is genetically distinct from P. micans. Cytologically, P. texanum has two parietal chlo-roplasts, each with a compound, interlamellar pyrenoid, trichocysts, fibrous vesicles that resemble mucocysts, pusules, V- to U-shaped posterior nucleus, golgi, and tubular mitochondria. No genetic difference was found between the two varieties in the five genes examined. Phylogenetic analysis of the SSU, LSU, and ITS ribosomal regions place P. texanum sp. nov. as a sister group to P. micans. One isolate of P. texanum var. texanum produces okadaic acid.

PHYLOGENETIC ANALYSIS OF BRACHIDINIUM CAPITATUM (DINOPHYCEAE) FROM THE GULF OF MEXICO INDICATES MEMBERSHIP IN THE KARENIACEAE(1).

Brachidinium capitatum F. J. R. Taylor, typically considered a rare oceanic dinoflagellate, and one which has not been cultured, was observed at elevated abundances (up to 65 cells · mL(-1) ) at a coastal station in the western Gulf of Mexico in the fall of 2007. Continuous data from the Imaging FlowCytobot (IFCB) provided cell images that documented the bloom during 3 weeks in early November. Guided by IFCB observations, field collection permitted phylogenetic analysis and evaluation of the relationship between Brachidinium and Karenia. Sequences from SSU, LSU, internal transcribed spacer (ITS), and cox1 regions for B. capitatum were compared with five other species of Karenia; all B. capitatum sequences were unique but supported its placement within the Kareniaceae. From a total of 71,487 images, data on the timing and frequency of dividing cells was also obtained for B. capitatum, allowing the rate of division for B. capitatum to be estimated. The maximum daily growth rate estimate was 0.22 d(-1) . Images showed a range in morphological variability, with the position of the four major processes highly variable. The combination of morphological features similar to the genus Karenia and a phylogenetic analysis placing B. capitatum in the Karenia clade leads us to propose moving the genus Brachidinium into the Kareniaceae. However, the lack of agreement among individual gene phylogenies suggests that the inclusion of different genes and more members of the genus Karenia are necessary before a final determination regarding the validity of the genus Brachidinium can be made.