Intelligent dissolved oxygen management in aquaculture ponds

In aquaculture, the success of a production cycle depends on a complex balance of factors, but none is as vital and dynamic as dissolved oxygen (DO) in the water. Maintaining stable DO concentrations is crucial to prevent hypoxia, a condition that can stress fish, affect their growth, suppress their immune system, and, in severe cases, cause mass mortalities with devastating economic losses.

Historical events, such as fish kills in the Philippines or the loss of millions of salmon in Chile due to harmful algal blooms, underscore the severe consequences of poor DO management. Fortunately, advancements in sensor technology, artificial intelligence (AI), and the Internet of Things (IoT) are revolutionizing how we monitor and predict oxygen levels.

A study by researchers at Florida Atlantic University, published in the journal Aquacultural Engineering, details the findings of a recent scientific review exploring the three pillars of modern DO management: the causal factors that govern it, the predictive models to anticipate its changes, and the smart monitoring technologies to measure it in real-time.

Key findings

• Dissolved oxygen (DO) is the most critical parameter governing the health of farmed species and the sustainability of aquaculture operations.
• Factors such as temperature, photosynthesis, organism respiration, and the decomposition of organic matter directly influence DO availability.
• Artificial intelligence (AI), through deep learning models, offers powerful tools to predict DO fluctuations with high accuracy, enabling proactive management.
• Optical sensors, integrated with Internet of Things (IoT) technology, represent the forefront of real-time monitoring, as they are more stable and require less maintenance than traditional electrochemical methods.
• The integration of predictive models and real-time monitoring systems is the most promising strategy to automate aeration control and transform reactive management into a predictive one.

What factors determine oxygen levels in the pond?

The DO concentration in a pond is not static; it is the result of a complex interplay of physical, chemical, and biological processes. To better understand this, the study classifies these factors into two main groups.

Direct factors: Processes that add or remove oxygen

These are the processes that directly add oxygen to or remove it from the water.

• Aeration: This is the transfer of oxygen from the atmosphere to the water’s surface. It can occur naturally through wind or be induced artificially with paddlewheel aerators, which are especially crucial at night when photosynthesis ceases.
• Photosynthesis: Performed by phytoplankton and aquatic plants, this is the primary source of oxygen production during the day, releasing it as a byproduct of converting sunlight into energy.
• Respiration: All organisms in the pond (fish, invertebrates, plankton, and microorganisms) constantly consume oxygen for their metabolic processes. The balance between total photosynthesis and respiration determines the net DO dynamics.
• Decomposition and Nitrification: The breakdown of organic matter (uneaten feed, feces, dead plankton) by aerobic bacteria is a major oxygen sink, particularly at the bottom of the pond. Likewise, the nitrification process, which converts toxic ammonia into nitrate, also consumes significant amounts of DO.

Indirect factors: Conditions that shape the environment

These factors do not add or remove oxygen themselves but modify the conditions affecting the direct processes.

• Temperature: This is perhaps the most critical factor. As water temperature rises, its capacity to hold dissolved oxygen decreases. Simultaneously, the metabolism of organisms accelerates, increasing their oxygen demand.
• Thermal Stratification: Solar radiation heats the water’s surface, creating layers of different temperatures that limit vertical mixing. This can leave deeper layers with low oxygen levels, as decomposition consumes available DO without it being replenished from the surface.
• pH: pH levels influence key biological and chemical processes. For instance, a higher pH increases the toxicity of ammonia, causing respiratory stress in fish and forcing them to consume more oxygen to detoxify.
• Turbidity: An excess of suspended particles (clay, organic matter) limits light penetration, reducing the photosynthetic capacity of phytoplankton and, consequently, oxygen production.

Predicting the future of oxygen: From statistics to artificial intelligence

The complexity of these interactions makes DO prediction a challenge. Historically, mechanistic models based on physical and chemical principles were used, but their application in real-world systems is often costly and limited.

The Power of Machine Learning and Deep Learning

The revolution has arrived with data-driven models. These AI algorithms do not need to explicitly know all the physical laws; instead, they “learn” hidden patterns from large volumes of historical data.

• Models like LSTM (Long Short-Term Memory) and GRU (Gated Recurrent Units) are particularly effective at analyzing time-series data (such as hourly DO measurements) and capturing short-term and long-term dependencies.
• Hybrid models that combine different architectures (such as CNNs for spatial patterns and LSTMs for temporal ones) are proving to be highly effective.
• One of the most promising frontiers is Physics-Informed Neural Networks (PINNs). These integrate fundamental physical laws (like Henry’s Law on gas solubility) into the AI’s learning process. This allows the models to be more accurate and robust, even in scenarios with scarce data.

Intelligent Monitoring: The Eyes and Ears in the Pond

For predictive models to work, they need high-quality, real-time data. This is where intelligent monitoring systems play a crucial role.

The Technology Behind Dissolved Oxygen Sensors

DO measurement relies primarily on two technologies:

• Electrochemical Sensors: These are the more traditional type. They work by measuring an electrical current generated by the reduction of oxygen at an electrode. Although inexpensive, they require regular maintenance (changing membranes and electrolytes), and their accuracy can be affected by water flow and the presence of other gases.
• Optical Sensors: These represent the most advanced technology. They operate on the principle of “fluorescence quenching.” A luminescent material is excited with light, and the presence of oxygen reduces the intensity and duration of its fluorescence. These sensors do not consume oxygen, are very stable long-term, require minimal maintenance, and are not affected by gases like H₂S, making them ideal for continuous field deployment.

The Internet of Things (IoT) revolution in aquaculture

The true transformation comes from connecting these sensors via the Internet of Things (IoT). Sensor networks deployed in the pond continuously collect data on DO, temperature, pH, and other parameters. This information is transmitted wirelessly (using protocols like LoRaWAN or 5G) to a cloud platform.

There, AI models analyze the data in real-time, predict DO trends for the coming hours, and, if they detect a risk of hypoxia, can automatically activate aeration systems or send an alert to the producer.

Conclusion: Towards a predictive and resilient aquaculture

The effective management of dissolved oxygen no longer depends solely on experience and manual measurements. The integration of a deep understanding of the factors that affect it, the power of AI-based predictive models, and the reliability of smart monitoring with optical sensors and IoT is ushering in a new era in aquaculture.

This integrated approach allows a shift from reactive management (acting after a problem has occurred) to a proactive and predictive one, preventing critical events before they happen. Adopting these technologies not only ensures the welfare of the farmed species but also optimizes the use of resources like energy, reduces risks, and fosters more sustainable, efficient, and profitable operations.

Contact
Minghao Yu
Department of Electrical Engineering and Computer Science, Florida Atlantic University
Boca Raton, 33431, FL, USA
Email: myu2024@fau.edu

Reference
Yu, M., Feng, Y., Ouyang, B., Wills, P. S., & Tang, Y. (2026). Dissolved oxygen in aquaculture ponds: Causal factors, predictive modeling, and intelligent monitoring. Aquacultural Engineering, 112, 102634. https://doi.org/10.1016/j.aquaeng.2025.102634

Milthon Lujan

Editor at the digital magazine AquaHoy. He holds a degree in Aquaculture Biology from the National University of Santa (UNS) and a Master’s degree in Science and Innovation Management from the Polytechnic University of Valencia, with postgraduate diplomas in Business Innovation and Innovation Management. He possesses extensive experience in the aquaculture and fisheries sector, having led the Fisheries Innovation Unit of the National Program for Innovation in Fisheries and Aquaculture (PNIPA). He has served as a senior consultant in technology watch, an innovation project formulator and advisor, and a lecturer at UNS. He is a member of the Peruvian College of Biologists and was recognized by the World Aquaculture Society (WAS) in 2016 for his contribution to aquaculture.