Thursday, April 23rd, 2026
Marine ecological models have improved enormously over the past two decades. We can now simulate ocean currents, track nutrient cycling, and predict species distribution shifts under climate scenarios with impressive precision. Satellite remote sensing delivers terabytes of environmental data every day. Autonomous underwater vehicles map the seafloor in detail that would have been unthinkable a generation ago.
And yet, when it comes to understanding what actually happens to marine ecosystems, not in theory, but in practice, there is a conspicuous gap. Most models treat human activity as an external forcing variable, something static to be plugged in as a boundary condition. The fishing fleet operates at a fixed intensity. Tourism pressure is a seasonal constant. Coastal development follows a predetermined trajectory.
This is, to put it plainly, not how people work.
The ocean does not exist in isolation from the communities that depend on it. Fishers respond to regulation, fuel prices, weather forecasts, market demand, and word of mouth. Tourism operators shift their routes based on social media trends and customer reviews. Coastal planners weigh short-term economic returns against uncertain long-term environmental costs. These decisions, made by thousands of individuals every day, collectively shape the pressures that marine ecosystems face.
Ecological modellers have known this for a long time. Elinor Ostrom’s work on socio-ecological systems, which contributed to her 2009 Nobel Prize in Economic Sciences, established that natural resource outcomes depend fundamentally on governance structures, social norms, and individual decision-making, not just biophysical processes. In marine contexts, this insight has been reinforced repeatedly. A 2019 paper in Ecology and Society by Schlüter et al. argued that failing to account for human behavioural dynamics in ecosystem models leads to systematically biased predictions, particularly for fisheries management.
Yet integration of social and ecological data in marine models remains the exception rather than the rule. A review by Elsawah et al. (2020) in Environmental Modelling & Software found that the majority of socio-environmental models still treat human behaviour simplistically, often as rational optimisation or fixed rules, rather than capturing the adaptive, sometimes contradictory ways people actually make decisions.
The difficulty is not simply one of willingness. There are genuine structural reasons why social data lags behind ecological data in marine science.
Scale mismatches. Ecological data often covers large spatial extents, satellite imagery, oceanographic transects, basin-wide surveys. Social data, by contrast, tends to be local and context-specific. A fishing community in the Aegean operates under entirely different economic pressures, cultural norms, and regulatory frameworks than one in the North Sea. Aggregating social data across these contexts without losing the detail that makes it useful is a persistent methodological challenge.
Temporal resolution. Environmental sensors can record measurements continuously. Social surveys, interviews, and economic datasets are typically collected at intervals, annually, seasonally, or as one-off research exercises. This mismatch in temporal resolution makes it difficult to couple social and ecological dynamics in a meaningful way. Fishing effort might shift dramatically within a single season in response to a regulatory change or a sudden market shift, but the social data to capture that response often arrives months or years later.
Ethical and privacy constraints. Ecological data is, for the most part, ethically uncomplicated to collect and share. Social data is not. Information about individual fishing behaviour, income, or livelihood strategies involves human subjects and is governed by data protection regulations such as the EU’s General Data Protection Regulation (GDPR). Anonymisation, consent, and data governance all add layers of complexity that ecological data simply does not face. These are not obstacles to be wished away, they are important protections, but they do slow the pace at which social data becomes available for modelling.
Disciplinary silos. Marine ecology, fisheries economics, sociology, and political science each have their own methodological traditions, publication norms, and professional networks. Genuinely interdisciplinary collaboration is harder than it sounds. As Kittinger et al. (2013) noted in Current Opinion in Environmental Sustainability, the institutional structures of academic research, from funding mechanisms to journal review processes, tend to reward disciplinary depth rather than cross-disciplinary integration. Building models that meaningfully combine social and ecological dynamics requires teams that span these divides, which takes time, trust, and institutional support.
The consequences of ignoring social dynamics in marine models are not merely academic. They shape real management decisions.
Consider fisheries management. Models that assume constant fishing effort may accurately predict stock trajectories under stable conditions, but they fail when conditions change, a new subsidy, a trade disruption, a sudden influx of vessels from another region. Fulton et al. (2011), working with the Atlantis ecosystem model, demonstrated that including human behavioural responses to management interventions produced substantially different and more realistic predictions than models treating fishing effort as fixed. Management strategies that looked optimal under static assumptions sometimes performed poorly when fishers adapted their behaviour.
Similar patterns emerge in coastal tourism. Models that treat visitor numbers as a simple function of season and weather overlook the social amplification effects of platforms like Instagram and TripAdvisor, which can concentrate pressure on specific sites far beyond what historical patterns would predict.
Marine spatial planning faces the same challenge. Designating a marine protected area is a biological and ecological decision, but its success depends on compliance, enforcement capacity, alternative livelihood availability, and community buy-in, all of which are social variables.
There is no single solution to the social data gap. But there are promising directions.
Advances in agent-based modelling allow researchers to simulate individual decision-making within ecological systems, representing fishers, tourists, and managers as adaptive agents rather than fixed parameters. These models are computationally demanding and require careful calibration against real behavioural data, but they offer a way to capture feedbacks between human decisions and ecological outcomes.
Participatory modelling approaches, where stakeholders are involved in the model-building process itself, can help bridge the gap between formal models and local knowledge. Projects that bring together fishers, marine scientists, economists, and policymakers around a shared modelling framework tend to produce results that are more credible and more useful to decision-makers.
Standardised social data collection protocols are also gaining traction. The Ocean Best Practices System, hosted by the International Oceanographic Data and Information Exchange, has begun to include social science methodologies alongside its established physical and biological standards. This is a small but significant step toward making social data as routine a component of marine research as temperature or chlorophyll measurements.
Initiatives like EcoTwin, which aim to build digital twins of marine ecosystems, face this challenge directly. A digital twin that captures only the biophysical environment is useful but incomplete. The real test is whether such systems can incorporate the messy, variable, context-dependent reality of human behaviour, and update dynamically as that behaviour changes.
Marine science has made extraordinary progress in understanding the ocean as a physical and biological system. The next frontier is understanding it as a social one. That means investing in social data infrastructure with the same seriousness we bring to oceanographic observation networks. It means building research teams that span disciplines, not just on paper but in practice. And it means accepting that some of the most important variables in marine ecosystem models are not measured in degrees Celsius or milligrams per litre, but in the decisions people make every day.
The data gap is real. Closing it will take sustained effort, institutional change, and a willingness to work across boundaries that have been comfortable for too long. But the alternative, continuing to model the ocean as if people were not part of it, is no longer good enough.