Safeguarding Australia's Research Outputs Through Long-Term Stewardship

Research data is accumulating faster than at any previous point in history, and Australian institutions are wrestling with a question their predecessors rarely had to face in such acute form. Once a project finishes, who keeps the data alive? Where does it live? Who pays for its continued care? And how do we make sure that a graduate student in Hobart today can still open a dataset collected by a field ecologist on the Wet Tropics coast twenty years from now?

The shift in how Australia's research sector talks about its outputs reflects this pressure. Conversations at the University of Melbourne, the Australian National University in Canberra, James Cook University in Townsville and smaller regional campuses increasingly treat datasets, code and accompanying documentation as first-class scholarly products. Funding bodies such as the Australian Research Council and the National Health and Medical Research Council have moved in step, demanding evidence that data will be managed responsibly and remain accessible beyond the grant period.

Stewardship is the term now used to describe this ongoing responsibility. It sits alongside storage and archiving but reaches further, since it covers the active curation, contextual preservation, controlled access and metadata enrichment that keep data meaningful over decades. The remainder of this article looks at what long-term stewardship involves, the policy setting in which Australian researchers operate, the national infrastructure that supports them, and the practical decisions that determine whether a dataset will still be readable, citable and useful in 2040.

Foundations of Data Stewardship

Stewardship is best understood as a continuum that runs through the entire life of a research project and beyond. The traditional research data lifecycle includes planning, collection, processing, analysis, sharing and preservation, and stewardship is the thread that holds each stage together. It begins at the moment a project is conceived, with decisions about file formats, naming conventions and metadata, and continues long after the principal investigator has moved on to the next grant.

Three distinctions matter. Stewardship is not the same as backup. A backup protects against accidental loss on a single workstation or server, while stewardship contemplates decades of institutional memory. Stewardship is also broader than open access. Some of the most carefully stewarded data in Australia, including health records and information about First Nations communities, is shared under controlled conditions. Stewardship is finally more than archiving. An archive may simply hold files untouched; a stewardship programme keeps those files discoverable, documented and, where appropriate, transformed into formats that survive technological change.

The practical elements include repository selection, persistent identifier assignment, licensing, sensitivity review and the production of rich contextual metadata. A dataset deposited in a trustworthy repository such as the Australian Data Archive in Canberra or the Tropical Data Hub at James Cook University comes with a documented chain of custody that the research community, funders and the public can rely on.

Australia's Policy and Legal Setting

Australian research operates inside a layered policy environment that shapes how stewardship is practised. The Australian Code for the Responsible Conduct of Research sets out expectations that data be retained for sufficient periods to allow verification and reuse, with minimum retention often set at five years and significantly longer where the data supports ongoing care, environmental monitoring or longitudinal studies. The Code also obliges researchers to ensure, where possible, that data remains accessible for further research.

Parallel obligations arise from the Privacy Act 1988 and its Notifiable Data Breaches scheme, which constrain what kinds of personal information can be shared and require that institutions report serious incidents. For health and social research this has produced a working culture in which a data management plan is treated as a privacy instrument, not merely an administrative one. The Office of the Australian Information Commissioner has also issued guidance on sharing data with overseas collaborators, which has practical consequences for projects with partners in Indonesia, Papua New Guinea and the Pacific.

A growing number of grant holders must now publish a data availability statement, lodge a data management plan at the proposal stage and use FAIR-aligned repositories. The University of Sydney, the University of Queensland and Monash University have all introduced internal policies that mirror these external requirements, which means that stewardship has effectively become a condition of doing research rather than a discretionary extra.

National Infrastructure and Trustworthy Repositories

A national layer of research data infrastructure has been built up over the past two decades, largely through the National Collaborative Research Infrastructure Strategy. The Australian Research Data Commons coordinates much of this activity from Melbourne, working with domain-specific facilities such as the Australian Bureau of Statistics' DataLab, the Terrestrial Ecosystem Research Network and the Australian Ocean Data Network.

The comparison table below gives a sense of the choices facing Australian researchers when they decide where a dataset will live.

Repository type	Typical location	Strengths	Limitations
Institutional repository	Hosted by the depositing university	Familiar to local researchers; aligns with grant requirements; supports preservation by the institution	Variable stewardship maturity; may not offer rich domain metadata
Domain or national repository	Hosted via NCRIS-funded infrastructure or specialist centres (e.g. Australian Data Archive, AODN)	Trusted core standards; persistent identifiers; strong domain metadata	May require more time to deposit; topic-bound
Generalist commercial or international repository	Hosted overseas (e.g. Zenodo, Dryad)	Easy to use; citable; broad uptake	Data sovereignty questions for sensitive material; uneven long-term guarantees
Bespoke project database	Built by individual researchers	Tailored to study needs	High risk of loss; no commitment to long-term curation

The most durable strategy is often a layered one. A project might keep working copies in a lab or research group server during the active phase, then deposit final versions in a domain repository affiliated with the Australian Research Data Commons, while retaining a descriptive record in the institutional repository at the institution. This redundant approach protects against the disappearance of any single host organisation.

Putting the FAIR Principles to Work

The FAIR principles, which call for data to be findable, accessible, interoperable and reusable, have become the operational language of stewardship in Australian universities. A findable dataset carries a persistent identifier such as a DOI, is indexed by a search service and is described with rich metadata. An accessible dataset can be retrieved by its identifier through a standard protocol, with authentication where needed. Interoperability means that the dataset speaks a shared vocabulary and uses formats that other systems can parse. Reusability relies on clear licensing, documented provenance and rich descriptive metadata.

Australian researchers have been quick to translate these words into practice. The Australian Antarctic Division in Hobart, for instance, maintains a multidisciplinary data collection that combines satellite observations, oceanographic profiles and biodiversity records, all of which are issued with DOIs and accompanied by metadata that complies with international standards. Similarly, the Atlas of Living Australia aggregates occurrence records from museums, herbaria and citizen science projects into a single searchable resource with stable citations.

Practical steps for a researcher preparing to deposit data include assigning a clear license, choosing repositories certified against recognised frameworks such as CoreTrustSeal, drafting a concise data dictionary, and adding machine-readable metadata at the point of creation rather than at the end of the project. These habits make stewardship easier to sustain and reduce the work required when a student or postdoc finally sits down to publish the dataset.

Stewardship Across Research Domains

Different disciplines ask different things of stewardship. In digital agriculture, where farmers and research agencies collaborate through trial sites stretching from the Darling Downs to the Ord River, the priority is high-resolution temporal data that can be revisited as new crop models are developed. In marine science, long time series from the Australian Institute of Marine Science or the Integrated Marine Observing System must be kept under certified archival conditions so that climate baselines are not eroded by the loss of a single disk array. In health research, the constraints of the Privacy Act 1988 often mean that data is shared through safe havens rather than repositories, yet the stewardship principles remain the same.

The most distinctive Australian conversation concerns Indigenous data. The CARE principles, which complement FAIR, were developed to articulate collective benefit, authority to control, responsibility and ethics in the governance of data drawn from First Nations communities. Research projects involving Aboriginal and Torres Strait Islander partners, whether in remote Western Australia or in suburban Melbourne, increasingly negotiate data agreements that specify not only storage and access but the conditions under which stories, knowledge and biological samples can be reused years into the future.

Domain challenges worth flagging include:

• The long-term cost of preserving high-volume environmental monitoring data such as radar, sonar and satellite imagery.
• The need to find and contextualise legacy data sets created before FAIR principles existed, including survey data from the 1970s and 1980s held on tapes or floppy disks.
• Sensitivity constraints in health and social research that prevent fully open sharing but demand equally rigorous documentation.
• Indigenous data sovereignty considerations that require community-led governance rather than top-down deposit.
• Cross-border data flows with neighbouring Pacific nations that raise both technical and diplomatic considerations.

People, Skills and the Question of Funding

Repositories and policies matter, but the actual work of stewardship is carried out by people. Australian universities have invested in a growing cohort of data stewards, data managers and research data librarians, often based in library and IT services and working closely with researchers across faculties. These roles involve negotiating data management plans, advising on repositories, performing quality checks, depositing datasets, training graduate students and acting as bridges between research offices, ethics committees and information governance teams.

The sustainability question is harder than the technical one. Funding for repositories often comes from project grants or short-term infrastructure investments, while the obligation to preserve data is open-ended. The Australian Research Data Commons has worked to professionalise the sector through training programmes, certification pathways and shared service models that distribute the cost across institutions. Even so, a dataset deposited today may need curation, format migration, metadata updates or sensitivity review for thirty years or more, and the institutional commitment that endures across leadership changes is the strongest predictor of survival.

Practical moves that strengthen long-term stewardship include:

• Building stewardship costs into grant budgets from the outset rather than treating them as institutional overhead.
• Appointing named data stewards with formally described responsibilities, supported by professional development.
• Using certified repositories whose governance outlasts the originating project team.
• Reviewing legacy datasets regularly to identify those at risk of becoming unreadable.
• Documenting the chain of custody so that future researchers understand the context of every observation and measurement.

None of these steps is especially glamorous, and none will appear at the top of a journal press release. Taken together, they are what make it possible for a young researcher in Adelaide, Perth or Darwin in 2045 to find, interpret and build on work carried out in this decade, and that is the practical meaning of long-term stewardship.