Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Machine Learning Reveals Large-scale Impact of Posidonia Oceanica on Mediterranean Sea Water

Celio Trois, Luciana Didonet Del Fabro, Vladimir A. Baulin

TL;DR

The paper investigates how the Mediterranean seagrass Posidonia oceanica modulates seawater biogeochemistry by linking precise meadow locations to a broad set of biogeochemical and physical variables. It leverages an augmented dataset of 174 features and supervised learning to identify indirect indicators of meadow presence, achieving high localization accuracy especially after adding depth, temperature, salinity, and transparency data. The study demonstrates robust relationships involving carbon-related fluxes and oxygen production, with a ROC-AUC near $0.96$ and a precision around $90\%$ for meadow detection. These findings support scalable mapping and conservation planning for blue-carbon ecosystems in the Mediterranean.

Abstract

Posidonia oceanica is a protected endemic seagrass of Mediterranean sea that fosters biodiversity, stores carbon, releases oxygen, and provides habitat to numerous sea organisms. Leveraging augmented research, we collected a comprehensive dataset of 174 features compiled from diverse data sources. Through machine learning analysis, we discovered the existence of a robust correlation between the exact location of P. oceanica and water biogeochemical properties. The model's feature importance, showed that carbon-related variables as net biomass production and downward surface mass flux of carbon dioxide have their values altered in the areas with P. oceanica, which in turn can be used for indirect location of P. oceanica meadows. The study provides the evidence of the plant's ability to exert a global impact on the environment and underscores the crucial role of this plant in sea ecosystems, emphasizing the need for its conservation and management.

Machine Learning Reveals Large-scale Impact of Posidonia Oceanica on Mediterranean Sea Water

TL;DR

The paper investigates how the Mediterranean seagrass Posidonia oceanica modulates seawater biogeochemistry by linking precise meadow locations to a broad set of biogeochemical and physical variables. It leverages an augmented dataset of 174 features and supervised learning to identify indirect indicators of meadow presence, achieving high localization accuracy especially after adding depth, temperature, salinity, and transparency data. The study demonstrates robust relationships involving carbon-related fluxes and oxygen production, with a ROC-AUC near and a precision around for meadow detection. These findings support scalable mapping and conservation planning for blue-carbon ecosystems in the Mediterranean.

Abstract

Posidonia oceanica is a protected endemic seagrass of Mediterranean sea that fosters biodiversity, stores carbon, releases oxygen, and provides habitat to numerous sea organisms. Leveraging augmented research, we collected a comprehensive dataset of 174 features compiled from diverse data sources. Through machine learning analysis, we discovered the existence of a robust correlation between the exact location of P. oceanica and water biogeochemical properties. The model's feature importance, showed that carbon-related variables as net biomass production and downward surface mass flux of carbon dioxide have their values altered in the areas with P. oceanica, which in turn can be used for indirect location of P. oceanica meadows. The study provides the evidence of the plant's ability to exert a global impact on the environment and underscores the crucial role of this plant in sea ecosystems, emphasizing the need for its conservation and management.
Paper Structure (20 sections, 4 figures, 5 tables)

This paper contains 20 sections, 4 figures, 5 tables.

Table of Contents

  1. Introduction
  2. Results
  3. Key Biogeochemical Variables
  4. Detecting P. oceanica with larger set of indirect indicators
  5. Visualizing the Predicted Locations
  6. Impact of P. oceanica on Sea Water
  7. Discussion
  8. Methods
  9. Dataset Creation
  10. Locations
  11. Random Points Generation
  12. Biogeochemical Variables
  13. Supplementary Variables
  14. Data Preprocessing
  15. Data Cleanup
  16. ...and 5 more sections

Figures (4)

  • Figure 1: Machine learning models performance and dimensionality reduction. a, Heat map showing the features correlations. b, Classification accuracy on reducing the correlated features. c, Classifiers accuracy on features with correlation $<=$ 0.8. d, Hierarchical clustering dendrogram.
  • Figure 2: Machine learning models performance evaluation on biogeochemical, depth, water temperature, salinity, and transparency, considering variables correlation $\le 0.8$. a, Classifiers accuracy. b, Confusion Matrix. c, Receiver Operating Characteristic (ROC) curve.
  • Figure 3: Map of studied area, highlighting predicted values. a, Generated data points near L’Ametlla de Mar, Catalonia. b, Pixel problem on predictions with biogeochemical variables. c, Predictions using biogeochemical, bathymetry, water temperature, salinity, and transparency variables. d, Zoom in on the P. oceanica grasslands of the city of Denia, Valencian community. e, L’Ametlla de Mar. f, Vilanova i la Geltrú, Catalonia. g, Alcudia Bay, north coast of Mallorca island. h, Almerimar, Andalucia. i, La Manga, Murcia. j, Colonia de San Jordi, southern part of Mallorca, Balearic Islands.
  • Figure 4: Random Forest feature importance. a, Using only biogeochemical data. b, Using biogeochemical, depth, water salinity, temperature, and transparency. c, KDE pairwise relationships; green refers to Posidonia data, while red indicates Non Posidonia.