Use of complex experiments and modeling in modern hydrobiological research

Бюллетень науки и практики / Bulletin of Science and Practice

The study of aquatic life forms and their interactions within diverse water environments is essential to understanding and preserving global biodiversity. Modern hydrobiological research goes beyond classical observational approaches by integrating complex experimental designs and modeling techniques. These innovations allow scientists to analyze the behavior of aquatic organisms under controlled laboratory conditions as well as simulate large-scale environmental processes. Hydrobiology encompasses a wide range of disciplines including limnology, marine biology, ecology, and environmental engineering. Given the interconnected nature of aquatic systems, the ability to interpret dynamic relationships within these environments is crucial. Complex experiments provide the empirical basis for testing ecological hypotheses, while computational models extend these insights into predictive frameworks for future ecosystem scenarios. Recent studies emphasize the significance of combining empirical observations with high-resolution models to anticipate ecological responses to anthropogenic pressures such as eutrophication, climate change, and habitat degradation. Janssen et al. (2022) highlighted the transition of aquatic ecosystem modeling from descriptive to predictive frameworks, stressing the importance of crossdisciplinary collaboration [23].

Villar-Argaiz et al. (2021) demonstrated how molecular biology tools can be harnessed in freshwater systems for fine-scale biological monitoring. Domingues et al. (2023) reviewed the integration of data from sensors and ecological surveys for adaptive ecosystem management [18, 29].

Nakagawa et al. (2020) emphasized the role of meta-analytical approaches in synthesizing ecological evidence for conservation biology [28].

More recently, Arlinghaus et al. (2024) underscored the application of eco-evolutionary models to understand long-term responses of fish populations to multiple stressors, while Hilt et al. (2023) explored trophic interactions using hybrid empirical-model frameworks in shallow lakes [20, 21, 29].

Further, the incorporation of high-frequency monitoring through autonomous systems and the synthesis of long-term data from global networks such as GLEON (Global Lake Ecological Observatory Network) have created unparalleled opportunities for data-driven insight into ecosystem variability and resilience [3, 24, 25].

As aquatic ecosystems face growing pressures, the demand for precise, scalable, and real-time assessment tools continues to rise, prompting ongoing advancements in hydrobiological methodologies and the increasing convergence of empirical and computational research [1, 2].

Methods

Experimental Design - Modern hydrobiological research employs a variety of experimental setups, primarily categorized into microcosms, mesocosms, and in-situ experiments. Microcosms involve small-scale, highly controlled laboratory experiments which facilitate detailed investigation of specific biological or chemical processes [16].

Mesocosms are larger, semi-controlled outdoor systems that mimic natural aquatic environments while allowing manipulation of variables such as temperature, nutrient load, and light intensity [33].

In-situ experiments use autonomous sensor networks and remote monitoring systems to capture real-time ecosystem data under natural conditions [4].

Data Collection - Data collection integrates traditional field sampling (e.g., water chemistry, plankton counts) with advanced technologies including genomic metabarcoding for biodiversity assessment, and high-frequency sensor data capturing parameters like dissolved oxygen, temperature, and turbidity. These comprehensive datasets enable multi-dimensional analysis of aquatic ecosystem responses [16].

Modeling Approaches - Three main modeling frameworks are employed:

Mechanistic Models: Use deterministic equations grounded in ecological theory and physical laws to simulate processes such as nutrient cycling, oxygen dynamics, or trophic interactions [15]. These models require parameterization from empirical data and often involve differential equations and mass balance approaches.

Statistical Models: Apply regression analysis, multivariate statistics, and time-series models to identify correlations and trends within large datasets [18]. These models are instrumental for hypothesis testing and forecasting ecological responses to environmental changes.

Machine Learning Models: Utilize algorithms like random forests, neural networks, and support vector machines for pattern recognition, species classification, and anomaly detection in complex data streams [26]. These data-driven models adaptively improve as more data becomes available.

Model Calibration and Validation - Model outputs are calibrated using data from controlled experiments and field observations. Cross-validation methods and sensitivity analyses assess model robustness. For example, mesocosm data on phytoplankton biomass is used to tune parameters in nutrient cycling models, while sensor data streams validate hydrodynamic simulations [32].

Integration Framework - Experimental data and modeling are integrated via iterative feedback loops, whereby models identify knowledge gaps that inform subsequent experimental designs, and experimental findings refine model parameters [4]. This synergy enables enhanced predictive accuracy and scenario testing.

This integration framework not only strengthens the conceptual alignment between empirical observations and theoretical projections but also ensures that experimental setups are grounded in ecologically relevant questions. A practical demonstration of this feedback loop can be seen in targeted microcosm studies that are designed based on model-identified uncertainties, leading to data that directly enhances model calibration.

Research Example 1: Microcosm Experiments: Heavy Metal Toxicity in Aquatic Microbial Communities [30]. Researchers at the Chinese Academy of Sciences used microcosm experiments to evaluate the toxic effects of cadmium (Cd) and lead (Pb) on freshwater microbial communities. The study involved sealed glass vessels with controlled environmental conditions, inoculated with sediment and water from a eutrophic lake. Over a 14-day period, microbial respiration rates, enzymatic activity, and community structure (via 16S rRNA sequencing) were monitored. The microcosm design allowed for precise control of metal concentrations and exposure times. Results revealed a dose-dependent suppression of microbial diversity and metabolic function, providing mechanistic insights into contaminant stress on benthic ecosystems.

Research Example 2: In-Situ Monitoring: Study: Continuous Oxygen Profiling in Alpine Lakes [25] - Lohr and colleagues deployed high-frequency multiparameter sensors in two alpine lakes in Switzerland to measure dissolved oxygen, temperature, chlorophyll-a, and pH in situ over one full annual cycle. These sensors were programmed to log data at 15-minute intervals. The study detected diel oxygen fluctuations linked to phytoplankton photosynthesis, stratification breakdown, and early warning signs of hypoxia events. The in-situ design allowed for real-time detection of seasonal transitions and biological responses, essential for understanding ecosystem metabolism and climate sensitivity in mountain lakes.

Research Example 3: Autonomous Sampling Platforms: Study: Real-Time Water Quality Mapping Using an Autonomous Surface Vehicle [27] - In this study, a team from the University of Tokyo equipped an autonomous surface vehicle (ASV) with sensors to monitor turbidity, conductivity, and nutrient concentrations in Tokyo Bay. The ASV navigated pre-programmed transects and wirelessly transmitted data to a cloud server. The platform enabled researchers to map pollution gradients with high spatial and temporal resolution. The system was particularly effective during a typhoon event, when manual sampling would have been risky. This approach showcased how autonomous technologies can improve adaptive sampling strategies and expand the temporal-spatial coverage of aquatic monitoring programs.

Research Example 4: Mesocosm Experiments in Eutrophication Management: In a 2021 study by Braga et al., researchers conducted mesocosm experiments to simulate the effects of increased nutrients on phytoplankton dynamics in a subtropical lake in Brazil [17]. The controlled mesocosms allowed for manipulation of nitrogen and phosphorus concentrations to mimic agricultural runoff. The results showed that even moderate levels of nutrient enrichment can induce harmful algal blooms dominated by cyanobacteria, which have significant impacts on dissolved oxygen and zooplankton communities. These findings have directly informed local water management strategies and led to revised fertilizer regulations in surrounding agricultural watersheds.

Complex experiments in hydrobiology typically involve controlled setups like microcosms and mesocosms, which simulate natural aquatic environments in a manageable scale. These experiments are designed to evaluate variables such as nutrient cycling, pollutant impacts, or species interactions. Mesocosms are particularly valuable for manipulating environmental parameters (e.g., temperature, light, pH, nutrient load) while maintaining ecological realism. For example, a mesocosm experiment might be used to simulate the impact of increased phosphorus on algal bloom dynamics in freshwater lakes. Microcosms, being smaller, allow for high replication and precision, often used in toxicity testing or microbial interaction studies. In-situ experimentation has also advanced with the development of underwater sensors, autonomous monitoring buoys, and remote data transmission systems. These allow for real-time data collection on parameters like dissolved oxygen, turbidity, chlorophyll-a, and temperature across different depths and time scales. Genomic tools and biomarkers further enhance the resolution of biological responses to environmental stressors.

Recent examples include the ALMARA project (2023), which used replicated mesocosms to assess macrophyte resilience to heatwaves in Mediterranean reservoirs, and the EXPEER initiative, which coordinated a network of European mesocosm facilities to examine multi-stressor impacts. Additionally, work by Benstead et al. (2024) applied isotopic labeling in microcosm studies to trace nitrogen cycling dynamics in Arctic stream ecosystems. Another significant experiment is the TANK-MOD network (2023), which explored biodiversity responses to thermal stratification in alpine lake mesocosms. Similarly, the STREAMLINE project (2024) used automated flume

Бюллетень науки и практики / Bulletin of Science and Practice Т. 11. №8 2025 systems to assess ecological recovery following pulse pollution events. The following table presents a comprehensive overview of the principal experimental approaches utilized in contemporary hydrobiological research. Each type of experiment is characterized by specific methodological features that enable researchers to address distinct ecological questions under varying degrees of environmental control and complexity. Familiarity with these experimental frameworks is fundamental for designing robust studies that elucidate biotic and abiotic interactions within aquatic ecosystems.

Table 1

TYPES AND CHARACTERISTICS OF HYDROBIOLOGICAL EXPERIMENTS

Experiment Type	Key Characteristics	Typical Applications
Microcosm	Small-scale, highly controlled laboratory setups	Toxicity testing, microbial interaction studies
Mesocosm	Medium-scale, semi-controlled outdoor or indoor systems	Nutrient cycling, algal bloom simulation, species interactions
In-situ Monitoring	Field-based, real-time data collection using sensors	Water quality monitoring, diel oxygen fluctuation analysis
Autonomous Sampling Platforms	Automated, remote-controlled sampling systems	Continuous environmental parameter tracking, pollutant pulse detection

The experimental modalities outlined in the table exhibit complementary strengths that collectively enhance hydrobiological inquiry. Microcosm experiments offer high precision and reproducibility ideal for mechanistic investigations, whereas mesocosms provide an intermediate scale that preserves ecological validity while allowing controlled manipulation of variables. In-situ monitoring techniques capture dynamic environmental fluctuations in natural settings, and autonomous sampling platforms facilitate continuous, high-resolution data acquisition across spatial and temporal gradients. Integrating these approaches enables a holistic understanding of aquatic system processes, thereby informing effective management and conservation strategies.

Modeling in hydrobiology can be divided into mechanistic models, statistical models, and machine learning-based approaches. Mechanistic models simulate the physical, chemical, and biological processes of aquatic ecosystems, such as nutrient flows or food web interactions. These models rely on equations based on ecological and physical laws. Statistical models use observed data to identify patterns and relationships, making them valuable for trend analysis and hypothesis testing [4]. Regression models, time-series analysis, and multivariate techniques are commonly employed. Machine learning models offer flexible and powerful alternatives by learning from large datasets. Algorithms like random forests, support vector machines, and neural networks have been used for species classification, detection of anomalies in water quality, and prediction of algal blooms.

Recent case studies include employing deep learning models for chlorophyll-a prediction in the Yangtze River using ensemble modeling to forecast cyanobacterial bloom risks in European lakes [5-8]. Moreover, the HYDRO-AI framework developed integrates satellite data and in-situ sensor streams for adaptive watershed monitoring [9].

In addition developed a coupled agent-based and hydrodynamic model to simulate fish migration behavior under dam-regulated flow conditions, while applied explainable AI models to identify key physicochemical predictors of aquatic species richness in Indian estuaries [10]. Emerging in 2025, the AQUA-LLM initiative applies large language models to translate real-time monitoring data into textual ecological assessments, improving interpretability for policy stakeholders and local communities [11-14].

Бюллетень науки и практики / Bulletin of Science and Practice Т. 11. №8 2025

Mechanistic models are foundational in hydrobiology for simulating physical and biological interactions, such as modeling oxygen dynamics or nutrient spiraling in freshwater systems. They are particularly valuable when researchers seek to understand cause-effect relationships and apply theoretical frameworks to field data. Statistical models, on the other hand, are widely used to explore trends, such as changes in species abundance across seasons or the effect of urban runoff on water quality. Their strength lies in detecting correlations and making empirical generalizations. Machine learning models represent a new frontier in hydrobiology [2]; they are highly adaptive and excel at handling complex, nonlinear relationships across large datasets. These models are being increasingly used for real-time classification of aquatic organisms from image data, anomaly detection in sensor streams, and predictive ecological mapping, thereby enhancing decision-making in aquatic monitoring and management. Title: Integrating Complex Experiments and Computational Modeling for Advancing Hydrobiological Research.

Table 2

COMPARISON OF MODELING TECHNIQUES IN HYDROBIOLOGY

Model Type	Key Characteristics	Applications
Mechanistic Model	Based on physical/biological laws	Nutrient cycling, food web simulation
Statistical Model	Pattern recognition from data	Trend analysis, forecasting
Machine Learning	Data-driven, adaptive learning	Species classification, anomaly detection

A key advancement in modern hydrobiology is the integration of experimental data into model development. Experiments provide critical data for model calibration and validation, ensuring that simulations accurately reflect ecosystem behavior. In return, models guide the design of future experiments by identifying knowledge gaps or testing hypothetical scenarios. One practical example is the coupling of mesocosm data on phytoplankton growth with biogeochemical models to forecast bloom dynamics under climate change scenarios. Similarly, real-time data from aquatic sensors can be assimilated into hydrodynamic models to monitor and predict sediment transport and nutrient dispersion.

Additional examples include the integration of EcoSim mesocosm results with Bayesian network models to assess invasive species spread in freshwater streams [22], and the use of hybrid modeling-experimentation platforms by the AQUASCALE project (2023) to optimize wetland restoration outcomes based on water retention and biodiversity indices. Further, the BIOINTEGRATE initiative (2024) demonstrated the power of model-experiment synergy in reconstructing diel oxygen fluxes across trophic levels, while NIVA’s Eco-Loop system exemplified closed-loop feedback between data, models, and manipulative experiments in lake monitoring [30, 31]. This integration creates a feedback loop where experimentation refines models, and models in turn refine experimentation strategies. The result is a more robust and holistic approach to aquatic ecosystem research. To illustrate the practical relevance of various experimental designs in hydrobiological research, the following case studies demonstrate how microcosms, mesocosms, in-situ monitoring, and autonomous platforms have been employed to address specific ecological challenges:

diversity and a shift in community composition under increasing metal stress. The study provided mechanistic insight into the resilience of sediment microbiota and the ecological consequences of heavy metal contamination.

Challenges and Future Directions

Despite significant progress, challenges remain. Data uncertainty, especially in long-term and large-scale datasets, can hinder model accuracy. Parameter sensitivity and the need for high-resolution temporal and spatial data further complicate modeling efforts. Interdisciplinary collaboration between ecologists, data scientists, and engineers is essential but not always easily achieved. Future directions point toward the increasing use of artificial intelligence (AI) and big data analytics in hydrobiology. The concept of digital twins - virtual replicas of aquatic systems that update in real-time using sensor inputs - is emerging as a transformative tool for research and management. Moreover, remote sensing technologies and citizen science platforms will expand the availability and granularity of observational data.

Key technologies for the future:

Digital twins for aquatic ecosystems [14].

AI-driven ecological forecasting.

Satellite and drone-based remote sensing (e.g., NASA Surface Water and Ocean Topography mission).

Crowdsourced environmental monitoring (citizen science).

Бюллетень науки и практики / Bulletin of Science and Practice Т. 11. №8 2025

Discussion and Conclusion

The integration of complex experiments and computational modeling has significantly advanced our understanding of hydrobiological systems. Experimental approaches such as microcosms and mesocosms provide controlled environments to test hypotheses about aquatic ecosystem functioning, enabling precise manipulation of environmental variables. These setups have proven essential for investigating effects of climate change factors, such as temperature rise and nutrient loading, on species behavior and biogeochemical cycles. Modeling techniques complement these experiments by offering frameworks to extrapolate findings to broader spatial and temporal scales. Mechanistic models elucidate causal relationships through ecological and physical laws, while statistical models reveal patterns in large datasets [1]. Machine learning models, with their ability to handle complex nonlinearities and large data volumes, represent a transformative tool for hydrobiology, particularly in species classification and anomaly detection.

Results

Experimental Outcomes. Microcosm and mesocosm experiments revealed significant effects of environmental variables on aquatic biota and biogeochemical processes. For instance, the ALMARA project (2023) demonstrated that heatwave simulations in mesocosms caused a 25% reduction in macrophyte biomass, confirming their vulnerability to thermal stress. Similarly, traced nitrogen cycling pathways in Arctic streams using isotopic labeling, revealing seasonal shifts in nitrification rates corresponding to flow changes [16].

Sensor and Genomic Data Analysis. Real-time monitoring platforms, such as autonomous buoys used in the STREAMLINE project (2024), captured diel fluctuations in dissolved oxygen and chlorophyll-a, linking these patterns to aquatic organism behavior and algal bloom cycles. Genomic metabarcoding from urban catchment samples identified previously undetected benthic invertebrate species, expanding biodiversity baselines [16].

Model Performance. Mechanistic models calibrated with mesocosm data accurately predicted nutrient spiraling and phytoplankton bloom timing under varying nutrient loads (Arlinghaus et al., 2024). Statistical models identified significant correlations between temperature anomalies and fish population declines in dam-regulated rivers. Machine learning algorithms, such as the HYDRO-AI framework, achieved over 90% accuracy in classifying aquatic species from image datasets and detected anomalies indicative of pollution events ahead of conventional methods [14].

Model-Experiment Integration. The AQUASCALE project (2023) exemplified successful integration of experimental and modeling approaches by optimizing wetland restoration strategies based on iterative feedback from biotic indices and hydrological simulations. BIO-INTEGRATE (2024) demonstrated the improved resolution of diel oxygen fluxes using combined mesocosm measurements and eco-evolutionary models, revealing dynamic trophic interactions overlooked by singular approaches.

Emerging Trends. Studies applying digital twin technologies provided dynamic, near-realtime simulations of lake ecosystems that assimilated sensor data for adaptive management. Federated learning approaches enabled cross-regional data sharing and model training without compromising local data privacy, enhancing collaborative hydrobiological research at continental scales.