Elsevier

Harmful Algae

Volume 103, March 2021, 102007
Harmful Algae

Original Article
Identification of influencing factors of A. catenella bloom using machine learning and numerical simulation

https://doi.org/10.1016/j.hal.2021.102007Get rights and content

Highlights

  • Outbreak of A. catenella was found on Geoje Island, South Korea in 2017.

  • Influence factors to A. catenella were found by machine learing and physical model.

  • Coastal water of Geoje Island had weak advection and dispersion in 2017.

  • Growth of A. catenella was influenced by water temperature and PO4P.

  • Density of A. catenella accelerated with an increase in retention time.

Abstract

Alexandrium catenella (A. catenella) is a notorious algal species known to cause paralytic shellfish poisoning (PSP) in Korean coastal waters. There have been numerous studies on its temporal and spatial blooms in Korea. However, its bloom dynamics have not been fully understood because of the complexity in physical, chemical, and biological environments. This study aims to identify the factors that influence A. catenella blooms by applying a numerical model and machine learning. Intensive monitoring of A. catenella was conducted to investigate temporal variations in its population and its spatial distribution in the area with frequent occurrences of PSP bloom initiation. Moreover, a numerical model was built to analyze the ocean physical factors related to the bloom of A. catenella. Based on the information obtained from the monitored and simulated results, the decision tree (DT) method was applied to identify factors that caused the bloom. The outbreak of A. catenella was observed in the eastern coastal water of Geoje Island in 2017, recording a peak density of 4 × 104 (cell L−1). Retention time and particle scattering demonstrated that the physical force in 2017 was weaker than that in 2018, as shown by the smaller effects of advection and dispersion in 2017. The decision tree model showed that (1) water temperature below 17.21 °C was ideal for the growth of A. catenella, (2) phosphate influenced the growth of the species, and (3) cell density was accelerated with increasing retention time. The results from DT can contribute to the prediction of A. catenella blooms by determining the conditions that cause bloom initiation. Further, they can be used as a practical approach for mitigating HABs. Thus, machine learning and numerical simulation in this study can be a potential approach for effectively managing the bloom of A. catenella.

Introduction

Harmful algal blooms (HABs) tend to harm human health and the economy in affected areas (Anderson et al., 2012). As the geographical scale, frequency, and period of HABs have increased globally in recent decades, they have become a serious threat to aquatic ecosystems (Gobler et al., 2017; Hoagland et al., 2002; Weiher and Sen, 2006). In particular, Asian countries have frequently experienced HABs with respect to Alexandrium species (Han et al., 2016). Blooms of A. catenella, which is a dinoflagellate species known to cause paralytic shellfish poisoning (PSP), has been observed regularly in spring (Han et al., 1992; Ishikawa et al., 2014; Kim et al., 2002). In particular, the southeastern coastal areas of Korea have been reported as hotspots of toxic species (Han et al., 1992; Ishikawa et al., 2014; Kim et al., 2002; Shin et al., 2017).

Although numerous studies have been conducted to understand the mechanism and control of Alexandrium blooms, bloom dynamics have not been fully understood due to the complexities of physical, chemical, and biological environments (Anderson et al., 2012). McGillicuddy Jr (2010) demonstrated that the physical, chemical, and biological factors heterogeneously triggered the initiation, development, and ultimate demise of HABs. In particular, HABs are often associated with weather conditions (Brandenburg et al., 2017). Changes in hydrodynamics have been reported to cause rapid and massive development of HABs (Pettersson and Pozdnyakov, 2013). In addition, there also existed the interactive relationship between the environmental variables such as the water velocity affects the dispersion rate of nutrients and temperature stimulates the growth of cyst (Fischer, 1976; Pospelova et al., 2008). The typhoon and anthropogenic source can also change the trend of bloom (Ding et al., 2012; Gobler et al., 2017; Heisler et al., 2008; Zhu et al., 2014). Therefore, these complex process of HAB might impede management of it.

Given the limited resources, the modeling and machine learning approach could be an alternative to understanding and controlling harmful algae (Pyo et al., 2019; Ruardij et al., 2005; Yoshioka and Yaegashi, 2018). The numerical model of the ocean, along with its physicochemical background, has been used to simulate salinity, water temperature, and water elevation (Murtugudde and Busalacchi, 1998; Thompson et al., 2006). Salinity, water temperature and water elevation, and nutrients have been found to influence the outbreak of algal blooms (Armi et al., 2011; Bearon et al., 2006; Ralston et al., 2014). In particular, the numerical model can be useful to identify the temporal-spatial distribution of algae (Ralston et al., 2014). Several studies have been conducted to simulate HABs in the ocean. Pinto et al. (2016) used a particle-tracking model to predict abundance of HAB species. He et al. (2008) developed a numerical model to predict A. catenella (former A. fundynes) bloom in the western Gulf of Maine. Previous modeling studies have focused on the quantification of algae. In those studies, the numerical models were useful for generalizing the overall trends of algae populations. However, those models are limited in their ability to explain specific situations such as unexpected outbreaks and extinction of blooms due to the complex blooming mechanisms.

The limitations of the numerical models can be addressed by using a data-driven model, namely machine learning. Machine learning is capable of handling big data and can identify the most important and consequential features from a wide variety of variables (e.g., categorical and continuous variables) (Szegedy et al., 2015; Young et al., 2018). Machine learning (Chen and Manning, 2014; Litjens et al., 2017; Veselý et al., 2013; Young et al., 2018) has been applied in numerous fields such as image recognition, speech analysis, and biology. Among machine learning models, the DT model is suitable for recognizing influencing factors from the data because it can interpret a trained model by examining the tree of the model (Safavian and Landgrebe, 1991; Song and Ying, 2015). Kamikawaji et al. (2016) detected errors in water temperature and salinity observations using the DT model. Liu et al. (2015) estimated sea surface salinity in coastal waters using random forest. Although the model using machine learning showed the acceptable performance, most studies mainly focus on the simulations of water temperature, salinity, and water depth. In addition, few studies have applied machine learning to HABs.

Herein, we identify the factors influencing the bloom of Alexandrium species using a physical and machine learning model. A. catenella was selected as the target species; it is known to produce PSP and it frequently occurs during spring in South Korea (Ki et al., 2004; KI and HAN, 2007). The numerical model generated the temporal-spatial distribution of the ocean's environmental factors in Geoje Island, South Korea, which had frequently experienced outbreaks of A. catenella bloom. Machine learning estimated the extent of the bloom and analyzed the factors affecting it. The aims of this study were 1) to conduct temporal and spatial monitoring of A. catenella, 2) analyze the physical properties of the ocean using a numerical model, 3) estimate the extent of the A. catenella bloom using the DT model, and 4) identify the factors influencing the bloom of A. catenella by analyzing the tree model.

Section snippets

Study site

Our study implemented temporal and spatial monitoring to investigate the occurrence of A. catenella. The study area is located on the eastern coastal waters of Geoje Island in South Korea (Fig. 1). A spatial survey was conducted in the entire region near Geoje Island. Horizontal distributions of A. catenella were surveyed at 34 stations in March 2017 and 34 stations in April 2017; moreover, 24 stations were surveyed in March 2018 (Fig. 3). Temporal sampling was conducted intensively at a fixed

Intensive monitoring of A. catenella

An extreme bloom of A. catenella was recorded in late March of 1989 and in early May of 1997 in the Chinhae Bay of the southeastern coastal waters in South Korea, having a density greater than 10,000 cells L−1 (Lee, 2005). In this study, the maximum density in 2017 was 44,012 cells L−1 and the temporal variation of A. catenella in 2017 showed a similar seasonality of species occurrence, indicating a spring outbreak of A. catenella bloom in mid-April (Fig. 4a). However, there was absence of this

Conclusions

This study investigated the factors influencing the bloom of A. catenella. Intensive and spatial monitoring was conducted to investigate the temporal variation and spatial distribution of the species, respectively. Additionally, a numerical model was built to analyze the physical factors of the ocean that contributed to the bloom of A. catenella. Using the monitoring and simulation results, this study identified the hierarchical structure of the bloom. The major findings of this study are as

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgement

This research was supported by the Basic Core Technology Development Program for the Oceans and the Polar Regions of the National Research Foundation (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2016M1A5A1027457).

References (110)

  • K. Ichimi et al.

    The growth and cyst formation of a toxic dinoflagellate, Alexandrium tamarense, at low water temperatures in northeastern Japan

    J. Exp. Mar. Biol. Ecol.

    (2001)
  • S. Jeong et al.

    Salinity intrusion characteristics analysis using EFDC model in the downstream of Geum River

    J. Environ. Sci.

    (2010)
  • H.-c. Kim et al.

    Relationship between phytoplankton bloom and wind stress in the sub-polar frontal area of the Japan/East Sea

    J. Marine Syst.

    (2007)
  • Y.O. Kim et al.

    Tracking Alexandrium catenella from seed-bed to bloom on the southern coast of Korea

    Harmful Algae

    (2020)
  • J. Li et al.

    Application of machine learning methods to spatial interpolation of environmental variables

    Environ. Model Software

    (2011)
  • G. Litjens et al.

    A survey on deep learning in medical image analysis

    Med Image Anal

    (2017)
  • M. Liu et al.

    Multivariable integration method for estimating sea surface salinity in coastal waters from in situ data and remotely sensed data using random forest algorithm

    Comput Geosci

    (2015)
  • R. Martyr-Koller et al.

    Application of an unstructured 3D finite volume numerical model to flows and salinity dynamics in the San Francisco Bay-Delta

    Estuar Coast Shelf Sci

    (2017)
  • D. McGillicuddy

    Models of harmful algal blooms: conceptual, empirical, and numerical approaches

    J Marine Syst.

    (2010)
  • S. Nagai et al.

    Monitoring of the toxic dinoflagellate Alexandrium catenella in Osaka Bay, Japan using a massively parallel sequencing (MPS)-based technique

    Harmful Algae

    (2019)
  • L. Pinto et al.

    Modeling the transport pathways of harmful algal blooms in the Iberian coast

    Harmful Algae

    (2016)
  • V. Pospelova et al.

    Distribution of dinoflagellate cysts in surface sediments from the northeastern Pacific Ocean (43–25 N) in relation to sea-surface temperature, salinity, productivity and coastal upwelling

    Mar. Micropaleontol.

    (2008)
  • V. Pospelova et al.

    Organic-walled dinoflagellate cyst production, composition and flux from 1996 to 1998 in the central Strait of Georgia (BC, Canada): a sediment trap study

    Mar. Micropaleontol.

    (2010)
  • J. Pyo et al.

    A convolutional neural network regression for quantifying cyanobacteria using hyperspectral imagery

    Remote Sens Environ

    (2019)
  • F. Recknagel

    Applications of machine learning to ecological modelling

    Ecol Modell

    (2001)
  • P. Ruardij et al.

    Modeling the bloom dynamics of the polymorphic phytoplankter Phaeocystis globosa: impact of grazers and viruses

    Harmful Algae

    (2005)
  • J. Sheikh-Ahmad et al.

    ANN constitutive model for high strain-rate deformation of Al 7075-T6

    J. Materials Process Technol.

    (2007)
  • H.H. Shin et al.

    Which species, Alexandrium catenella (Group I) or A. pacificum (Group IV), is really responsible for past paralytic shellfish poisoning outbreaks in Jinhae-Masan Bay, Korea?

    Harmful Algae

    (2017)
  • T.J. Smayda

    Complexity in the eutrophication–harmful algal bloom relationship, with comment on the importance of grazing

    Harmful Algae

    (2008)
  • C.O. Tan et al.

    Modeling complex nonlinear responses of shallow lakes to fish and hydrology using artificial neural networks

    Ecol Modell

    (2006)
  • D. Tang et al.

    In situ and satellite observations of a harmful algal bloom and water condition at the Pearl River estuary in late autumn 1998

    Harmful Algae

    (2003)
  • M.C.R. Alavanja et al.

    Health effects of chronic pesticide exposure: cancer and neurotoxicity

    Annu Rev Publ Health

    (2004)
  • A. Altmann et al.

    Permutation importance: a corrected feature importance measure

    Bioinformatics

    (2010)
  • D.M. Anderson et al.

    Progress in understanding harmful algal blooms: paradigm shifts and new technologies for research, monitoring, and management

    Ann Rev Mar Sci

    (2012)
  • Z. Armi et al.

    Alexandrium catenella and Alexandrium tamarense in the North Lake of Tunis: bloom characteristics and the occurrence of paralytic shellfish toxin

    Afr. J. Aquat. Sci.

    (2011)
  • R. Bearon et al.

    Effects of salinity structure on swimming behavior and harmful algal bloom formation in Heterosigma akashiwo, a toxic raphidophyte

    Mar. Ecol. Prog. Ser.

    (2006)
  • P. Boyd et al.

    Physiological responses of a Southern Ocean diatom to complex future ocean conditions

    Nat Clim Chang

    (2016)
  • R. Caruana et al.

    An empirical comparison of supervised learning algorithms

  • D.-.S. CHANG et al.

    A study on paralytic shellfish poison of sea Mussel, Mytilus edulis-food poisoning accident in Gamchun Bay, Pusan, Korea, 1986

    Korean J. Fisheries Aquat. Sci.

    (1987)
  • F. Changyong et al.

    Log-transformation and its implications for data analysis

    Shanghai Arch Psychiatry

    (2014)
  • C. Chen et al.

    An unstructured grid, finite-volume, three-dimensional, primitive equations ocean model: application to coastal ocean and estuaries

    J. Atmospheric and Oceanic Technol.

    (2003)
  • D. Chen et al.

    A fast and accurate dependency parser using neural networks

  • Z. Dai et al.

    Unusual salinity conditions in the Yangtze Estuary in 2006: impacts of an extreme drought or of the Three Gorges Dam?

    Ambio

    (2011)
  • L.B. de Miranda et al.

    Fundamentals of Estuarine Physical Oceanography

    (2017)
  • Y. Ding et al.

    Effects of typhoon Morakot on a large shallow lake ecosystem

    Lake Taihu, China. Ecohydrology

    (2012)
  • M.A. Doblin et al.

    Comparative study of selenium requirements of three phytoplankton species: gymnodinium catenatum, Alexandrium minutum (Dinophyta) and Chaetoceros cf. tenuissimus (Bacillariophyta

    J. Plankton Res.

    (1999)
  • Y. Fang

    Asymptotic equivalence between cross-validations and Akaike information criteria in mixed-effects models

    J. Data Sci.

    (2011)
  • Y. Feng et al.

    Chesapeake Bay nitrogen fluxes derived from a land-estuarine ocean biogeochemical modeling system: model description, evaluation, and nitrogen budgets

    J. Geophys. Res. Biogeosci.

    (2015)
  • H.B. Fischer

    Mixing and dispersion in estuaries

    Annu Rev Fluid Mech

    (1976)
  • T. Fushiki

    Estimation of prediction error by using K-fold cross-validation

    Stat Comput

    (2011)
  • Cited by (8)

    • Formation mechanism and environmental drivers of Alexandrium catenella bloom events in the coastal waters of Qinhuangdao, China

      2022, Environmental Pollution
      Citation Excerpt :

      In the present study, the temperature increased from 4 to 11 °C between the start and end of the bloom event, which is relatively lower than the temperature ranges of 7–15 °C (Condie et al., 2019) and 10–13 °C (Álvarez et al., 2019) reported in previous studies. Also, this increase is significantly lower than the ideal temperature of 17 °C estimated for the growth of A. catenella via machine learning and numerical simulations (Baek et al., 2021). This indicates the uniqueness of the algal strain in the region.

    View all citing articles on Scopus
    View full text