Artificial Intelligence and Estuarine Biodiversity Research

How can AI support scientific research?

It can locate and appraise literature, organise and classify data, structure databases, handle geographic and photographic records, identify patterns and prepare reports. It speeds up the preparatory work — which is most of the time in any research project — without replacing scientific method, field observation or validation by people who know the subject.

What was done

Starting from a project on the biodiversity of a major Portuguese estuary, we explored where artificial intelligence adds value and where it should not go. Specifically, we used AI to:

• locate and appraise scientific literature on estuarine ecosystems;
• compare the estuary with others, from the United Kingdom to North Africa;
• investigate species richness, abundance and distribution;
• analyse the different definitions of a biodiversity hotspot;
• structure a collection and classification methodology;
• organise species by phylum, taxonomic group and habitat;
• build a database for species, coordinates, dates, projects and photographs;
• prepare analyses by location, by period and by habitat type;
• relate photographic records to ecological information;
• prepare future observation and research activities on the river.

The definition you choose determines the answer

There is no single definition of a biodiversity hotspot. Some rest on species richness, others on endemism, others on degree of threat or rarity. The same estuary can be a hotspot under one criterion and not under another.

Stating which criterion — and why — is part of the scientific work, not a footnote. It is also exactly where an AI tool used carelessly produces a confident, wrong answer: the question «is this a hotspot?» has no answer without the question before it.

Why compare estuaries at different latitudes

An estuary on its own does not tell you whether its numbers are high or low. Comparing with estuaries in the United Kingdom and North Africa gives context to observed richness and distribution, and helps separate what is characteristic of that site from what is characteristic of the ecosystem type.

A database, not a list

Biodiversity records are only worth having if they are comparable. The structure links species, coordinates, date, project and photographic record, and organises species by phylum, taxonomic group and habitat — so that later analysis by location, by period and by habitat type is possible.

It is the same principle we applied in the healthcare data audit: before any analysis comes the question of whether the data can carry the analysis.

Three dimensions, and it is the link between them that matters

The part you cannot do at a computer

BigLearn keeps a practical connection to the river, including a boat on a major Portuguese river that can support observation, data collection, environmental awareness work and applied research.

It is the physical component of the same capability: being able to leave the digital environment and use the river as a space for observation, experiment and applied learning. The activities are in preparation, and the data structure was designed to receive whatever comes back from them.

We use AI to accelerate research while preserving scientific method, field observation and expert validation.

Where this applies beyond biodiversity

The same set — literature review and appraisal, organising and classifying data, identifying patterns and hypotheses, building scientific databases, handling geographic and photographic records, preparing methodologies and protocols, producing reports and visualisations, environmental monitoring and scientific communication — applies to any research project with documentary and field work.