Using Machine Learning to Observe Abundance Patterns of the Dino flagellate Noctiluca scintillans in the Western English Channel Noctiluca scintillans is a Harmful Algal Bloom (HAB) species with a wide geographic distribution. It frequently blooms and causes negative impacts on marine ecosystems [1], although this species is not thought to be a toxin producer. As a result of their high biomass bloom-forming abilities, N. scintillans can potentially cause anoxic conditions and the generation of high levels of ammonia leading to localised die-offs of higher trophic levels [2]. We aim to develop a rapid detection of this species from flow cell-sorted images of plankton off the south-west coast of the UK. To that end, we have trained a Random Forest (RF) algorithm to i) rapidly and accurately quantify the abundance of Noctiluca in seawater samples of mixed plankton communities, ii) aid the examination of temporal patterns in Noctiluca abundance in the long-term, and iii) create a monitoring tool to show when cell counts approach bloom levels. Seawater samples taken between March and December 2018 at the L4 autonomous buoy in the Western English Channel were processed using dynamic imaging particle analysis (DIPA) performed using FlowCam (Fluid Imaging Technologies, USA). All particles ranging in size from 380 to 1500 μm were automatically imaged and enumerated by VisualSpreadsheet (the software accompanying FlowCam). A RF algorithm was then trained and tested on 32 particle features (for example, length, width, area, transparency etc. as measured by VisualSpreadsheet) in a mixed plankton community of 246,000 total particles; 167,000 of these particles had been manually identified as Noctiluca cells to give a True Count against which to test the model. One of the challenges of training the RF model was to encompass the ontogenetic changes seen in Noctiluca cells over their development and so training sets were compiled from cell samples taken on random dates throughout the year. Fig. 1 shows examples of some of the morphological and textural variety of Noctiluca that our RF model is tasked to classify. The resulting model was then deployed on the 2018 dataset to estimate Noctiluca abundance. RF model abundance and True Count were compared using linear regression with a significance level = 0.05. A strong positive relationship was found between the RF abundance and True Count, with the RF model almost exactly matching the True Count (p < 0.001, R2 = 1.000). It must be noted that the model is limited by FlowCams ability to identify individual particles in a sample which is likely not as accurate as manual counting with microscopy [3] A total of 33 plankton samples were collected between March and December 2018 comprising 1,008,084 particles in total, 540,499 of which were manually identified as Noctiluca from the FlowCam images. To observe the temporal pattern of Noctiluca abundance, the RF model was deployed on the 2018 datasets and compared with total particle counts (Fig. 2). Fig. 1. Examples of morphological and textural variety of Noctiluca scintillans 14 HARMFUL ALGAE NEWS NO. 64 / 2020 Harmful Algae News An IOC Newsletter on Toxic Algae and Algal Blooms No. 64 - June 2020 www.ioc-unesco.org/hab The IOC Taxonomic Reference List of Harmful Microalgae The creation of the IOC Taxonomic Reference List of Harmful Microalgae was first discussed 23 years ago at the Fourth Session of the ever the information presented in publications is not always easy to interpret. Toxic blooms often contain several species and studies on individual species are therefore required to determine the species responsible for the toxin production. In addition, species are sometimes difficult to identify considered this to be identical to the previously described P. mexicanum [9]. This statement makes little sense, and Loeblich et al in fact do not mention P. mexicanum in their article. Faust [8] illustrated material which resembled P. rhathymum using the radical arrangement of some of the trichocys Massive salmon mortalities during a Chrysochromulina leadbeateri bloom in Northern Norway From mid May to mid June 2019, fish farmers along the coast of Nordland and Troms, northern Norway, experienced sudden mortalities of caged salmon [1] (Fig. 1). These mortalities were assumed to be due to a blo Fish kill in numbers [2] 13 000 ton fish 7.5 mill salmon 80 mill EUR 14 companies Fig. 2. Map of Norway showing the area where the Chrysochromulina leadbeateri bloom occurred causing massive fish kills in May-June 2019 References 1. Fiskeridirektoratet 2019. https:// www.fiskeridir.no/Akvakult Mass mortality of marine invertebrates associated with the presence of yessotoxins in northern Chile Fig. 1. Map of the study area showing A) Chilean coast; B) Pabellón de Pica, Tarapacá Region C) Bahía Inglesa, Atacama Region; D) Puerto Aldea, Coquimbo Region During the austral summer of 2019, ma research is needed to determine the mechanism of action and the toxin effects on tissues and cells of the main affected species. Finally, there is a need to establish an educational plan to protect the public and avoid the consumption and commercialization of potentially toxic marine invertebrates. First Report of Heterocapsa minima (Dinophyceae) from Aotearoa/ New Zealand Small planktonic armoured dinoflagellates within the genus Heterocapsa are currently represented by 20 species with some having a world-wide distribution [1-2]. Blooms of some Heterocapsa species have been associated with fi Fig. 2. Maximum likelihood (ML) molecular phylogenetic tree of Heterocapsa minima and H. steinii strains isolated from coastal areas of New Zealand based on the LSU rDNA D1/ D2 sequences (20 sequences, 700 positions). The alignment site corresponded to the 69768 bp site of a sequence from H. steinii Yellow-green tides could become a recurrent event along the Ligurian coast (Italy) Fig. 1. A) Sestri Levante and Baia del Silenzio bay, characterized by shallow water and mixed rocky-sandy seabed containing the macrophyte, Posidonia oceanica (dark grey) and carpet-like matte (light grey) habitat. B First report of a high biomass bloom of Peridinium quadridentatum (F. Stein) Gert Hansen from the tropical Cochin estuary SW coast of India Fig. 1. Map showing the area of P. quadridentatum bloom in Cochin estuary, southwest coast of India Peridinium quadridentatum (F. Stein) Gert Hansen is a cos Table 1. Physico-chemical parameters of P. quadridentatum bloom area in Cochin Estuary Parameters Water Temperature (oC) Salinity (psu) Dissolved Oxygen (ml L-1) Nitrate (μmol L-1) Silicate (μmol L-1) Phosphate (μmol L-1) Chlorophyll a (mg m-3) 28 20 2.01 10.6 23.3 1.3 27.5 higher turbidity, lowe Are mesophotic seamounts reservoirs for potentially toxic dinoflagellates associated with Ciguatera poisoning? A case study from the SW Indian ocean (expédition La Pérouse, 2019) Fig. 1. Location map of La Pérouse seamount off La Réunion. In the western Indian Ocean, a Ciguatera Fish Poisoning (CF Using Machine Learning to Observe Abundance Patterns of the Dino flagellate Noctiluca scintillans in the Western English Channel Noctiluca scintillans is a Harmful Algal Bloom (HAB) species with a wide geographic distribution. It frequently blooms and causes negative impacts on marine ecosystems [1] References 1. Sarma VVSS et al 2019. Mar Pollut Bull 138: 428-436 2. Faust MA & RA Gulledge 2002. Contributions for the US National Herbarium 42: 1-144 3. Graham MD et al 2018. Limnol Oceanogr-Meth 16: 669-679 Author Christian Bamber, The Marine Biological Association, Citadel Hill, Plymouth PL1 2PB International validation and recognition of method for paralytic shellfish toxins in bivalve molluscs Food safety scientists from Cefas (UK) and Cawthron Institute (New Zealand) have led an international study over the past four years to gain international recognition for a new method to quantify pa Aotearoa/New Zealands nationally significant Cawthron Institute Culture Collection of Microalgae (CICCM) The CICCM is critical to international research projects and the 500 isolates of microalgae and cyanobacteria in the collection have been sourced from oceans, lakes and rivers in New Zealand, the The 11th EASTHAB Symposium and 4th Philippine HAB Conference The 11th EASTHAB Symposium and 4th Philippine HAB Conference were held back-to-back from December 11 13, 2019 at the Microtel Hotel, Puerto Princesa, Palawan, the Philippines. The theme was Advances in Harmful Algal Bloom Research, Monito Fig. 1. Participants in the 11th EASTHAB Symposium/4th Philippine HAB Conference Fig. 2. Department of Science and Technology secretary Fortunato dela Peña with the keynote, plenary speakers and local organizing committee leads. Left to right: Dr. Ichiro Imai, representative of the Vice Mayor of Pu The ICES-IOC Working Group on Harmful Algal Bloom Dynamics 2020 Meeting The International Council for the Exploration of the Sea (ICES) and the Intergovernmental Oceanographic Commission of UNESCO (IOC) have collaborated closely for 26 years, stimulating research on HAB population dynamics and monit Fig. 2. Shared day between ICES-IOC WGHABD, ICES-IOC WGBOSV and ICES WGITMO ences using molecular methods as well as results from studies in Arctic areas. These studies presented data on HAB and invasive non-native species present in Arctic waters. A USA study focused on the distribution, community Typical benthic habitat with macroalgae and limestone rubble observed at ~ 60m depth on La Pérouse seamount, Indian Ocean (Photo courtesy of L. Ballesta). Eds-in-chief Beatriz Reguera, IEO, Vigo, Spain Eilen Bresnan, Marine Scotland, UK Regional Editors Caribbean: Ernesto Mancera jemancerap@unal.e