Ocean Acidification

Ocean acidification is a global process characterized by a decrease in seawater pH and an increase in acidity due to the absorption of atmospheric carbon dioxide by the oceans. Since the Industrial Revolution, approximately 25 to 30 percent of the carbon dioxide emitted into the atmosphere by human activities has been absorbed by the oceans. This phenomenon, commonly referred to in scientific literature as the “evil twin of global warming” or the “other CO₂ problem,” poses critical threats to marine ecosystems and the human economies dependent on them.

Chemical Process and Fundamentals

The process of ocean acidification begins when carbon dioxide in the atmosphere comes into contact with the ocean surface and dissolves in seawater. Dissolved CO₂ reacts with water molecules to form carbonic acid, a weak acid. Carbonic acid, which has an unstable structure, rapidly dissociates into bicarbonate and hydrogen ions. The increase in the concentration of free hydrogen ions in seawater leads to a drop in pH and an increase in acidity. This chemical process also affects the saturation level of carbonate ions in seawater. ^【1】

Excess hydrogen ions in the environment tend to react with carbonate ions to form additional bicarbonate. This reduces the availability of biologically accessible carbonate ions necessary for the formation of calcium carbonate minerals such as calcite and aragonite. ^【2】 Historical data indicate that the average pH of surface ocean waters has decreased by approximately 0.1 units since the Industrial Revolution. ^【3】 Considering the logarithmic nature of the pH scale, this decline corresponds to an increase in acidity of approximately 26 to 30 percent.^【4】 The current rate of acidification is estimated to be ten times faster than any observed in the last 300 million years.^【5】

Marine Ecosystems and Biogeochemical Impacts

This rapid change in ocean chemistry has profound effects on the physiology of marine organisms. Particularly for “calcifying” organisms that rely on calcium carbonate to build shells and skeletons, the following harmful impacts have been observed:

• Difficulty in Shell Formation: Organisms such as oysters, mussels, corals, sea urchins, lobsters, and pteropods (sea butterflies) face increasing difficulty in accessing the building blocks required for shell formation, which reduces their growth rates.
• Dissolution and Weakening of Existing Structures: Increased acidity causes erosion of existing shells. Research on pteropods, a key food source for salmon and whales, has shown that under projected 2100 acidity levels, their shells can dissolve within 45 days. Additionally, modern plankton shells have been found to be 76 percent thinner compared to 19th-century samples.
• Energy Imbalance and Physiological Stress: Organisms must expend more energy to maintain their internal chemistry and repair their shells. This reduces the energy available for vital functions such as growth, reproduction, immune response, and predator avoidance, leading to overall health deterioration.
• Coral Reef Degradation: As coral skeleton density decreases, reefs become more vulnerable to storms and recovery after bleaching events slows down. This threatens the habitat of thousands of species dependent on reefs.
• Disruptions in the Food Chain: The decline of calcifying organisms creates a domino effect, impacting higher trophic levels such as salmon and whales.

Turkish Seas: Mediterranean and Black Sea Analysis

Turkish seas exhibit unique resilience to acidification due to their chemical composition and geographic positioning. The Mediterranean Sea, with its high alkalinity and saturation levels of calcite and aragonite throughout all depths, acts as a natural buffer against acidification caused by rising CO₂ levels, potentially slowing its onset. The Black Sea similarly exhibits high calcite saturation in surface waters. However, despite these natural protections, Turkish seas face intense anthropogenic threats. Increasing coastal populations, rapid urbanization, and industrialization, combined with acidic compounds entering via rivers and the atmosphere, have the potential to trigger localized acidification processes. This necessitates continuous monitoring of Turkish marine environments.

Socioeconomic Impacts and the Blue Economy

The blue economy encompasses ocean-based activities such as fisheries, coastal tourism, and logistics. Ocean acidification threatens these sectors in the following ways:

• Fisheries and Seafood: Losses in shellfish productivity and larval mortality result in economic damage. Studies project annual losses of hundreds of millions of dollars by 2060. A study using Turkish data from 1990 to 2019 revealed a direct relationship between greenhouse gas emissions and the structure of intra-industry trade in the fisheries sector, demonstrating that acidification can alter these trade dynamics.
• Coastal Tourism: Coastal tourism, which depends on healthy coral reefs, is at significant risk due to reef degradation. Scientific projections estimate that by 2100, global economic losses due to declining coral health could reach $87 billion annually.
• Food Security: Coastal communities that rely on the ocean for protein face income declines and food insecurity, particularly in tropical regions.

Machine Learning Methods for pH Prediction and Modeling

Machine learning algorithms used to forecast acidification trends analyze comprehensive datasets and identify variables influencing pH levels. Analyses have determined that the factors most strongly correlated with pH levels are: the contribution of fossil fuel combustion to global warming, direct energy use (gas), and time (year). During the modeling process, missing values in raw data are imputed using mean, median, or mode, and the dataset is split into 80 percent for training and 20 percent for testing. The main models tested in research and their performance levels are as follows:

• Extra Trees: Demonstrated the highest performance among 11 algorithms. Its most notable feature is reduced risk of overfitting through random splits and strong performance on small datasets.
• Other Ensemble Methods: Models such as Decision Trees, Gradient Boosting, and Random Forests outperform simple linear models.
• Hybrid Models: Some studies have achieved high accuracy using multivariate structures combining 25 percent Decision Trees and 75 percent Gradient Boosting.
• Low-Performance Models: Models such as Linear Regression, K-Nearest Neighbors, SVM, and Neural Networks have been found to have lower accuracy compared to ensemble methods.