Blog

Digitalizing herbaria to think better about biodiversity

Note: This post was originally written in Spanish, and this English translation was made with AI.

Digitizing a biological collection is not just a matter of turning physical objects into data. It is a way of amplifying the work of collectors, taxonomists, and curators. Each herbarium specimen contains material evidence that a species was present in a place, on a particular date, under certain circumstances. When that information remains only in a folder, a label, or a cabinet, its reach is limited. When it is digitized, it becomes available to a much wider audience.

Digitized specimens allow us to broaden our view of biodiversity. When records from multiple collections are integrated, we can map the known distribution range of a species. That description is not just a cartographic exercise. It is a criterion for assessing conservation status. Under standardized methodologies, the extent of a species’ distributional range is a relevant indicator of extinction risk.

Herbaria also allow us to ask temporal questions. A species collected at a site several decades ago and not recorded there again may be showing a sign of local extinction. Digitized historical data allow us to compare the past with the present, detect changes in distribution, and establish baselines for monitoring programs.

Biological collections are infrastructures for observing change. In a context of biodiversity crisis, climate change, and accelerated transformation of ecosystems, we need to know where species are, where they were before, and where they may be disappearing from. Digitized herbarium specimens make it possible to sustain that conversation with evidence.

But there is one point that cannot be omitted. The phrase “digitizing without curation is democratizing ignorance” captures this problem well. If we take data with errors, outdated names, doubtful identifications, or misassigned coordinates, and place them on an open platform, we are not necessarily democratizing knowledge. We may indeed be amplifying error.

This warning is especially important in the age of artificial intelligence. Sometimes digitization and artificial intelligence are discussed as if they could replace the work of specialists. I see it differently. The more capacity we gain to process large volumes of data, the more important rigorous review becomes. The speed of digital workflows can cause a taxonomic error, an unresolved synonymy, or a mistaken identification to propagate downstream into many later analyses.

In BIODATA, we have seen that digitizing herbaria requires continuous taxonomic and cataloguing work. The scientific names in datasets must be updated, and identifications must be evaluated. Discrepancies do not disappear because data are published on a platform. Often, they become more visible. That visibility is positive, but only if there is a community that, before publication, takes care of correcting, interpreting, and updating the information.

That is why the work of naturalists and botanists remains fundamental. Artificial intelligence can help measure, classify, extract information from images, recognize patterns, program analyses, and accelerate processes that once took much longer. But it cannot replace the body of knowledge that allows us to know whether a datum makes sense. Artificial intelligence needs high-quality data to be genuinely useful. If the input data are weak, incomplete, or incorrect, the results will be too, even if they are produced with great efficiency.

From my point of view, the technological development of biological collections depends on modernizing the systems of work. Systems where artificial intelligence helps accelerate tasks, but where curation, traceability, and expert knowledge remain central.

In Chile, this discussion opens a very large opportunity. We have valuable biological collections, a scientific community with deep knowledge of the country’s flora and fauna, and technological capacities that have grown significantly in recent years. If we manage to digitize, integrate, and curate that information properly, we could build a biodiversity database with known levels of uncertainty and progressively lower uncertainty, which would allow us to move toward a modern biodiversity management system.

The challenge is to connect these elements in a common infrastructure capable of transforming specimens, data, and expert interpretations into indicators and actionable knowledge. In that sense, digitizing herbaria is not merely a technical task. It is a way of organizing and amplifying the country’s biodiversity knowledge so that we can think better about its future.