The Digital Tools for Planning and Managing Forest Restoration webinar has now run twice, and the second session, held on 30 September 2026, drew more than 150 registrants. We were glad to join the programme for this second run, alongside colleagues from the Royal Botanic Gardens, Kew, the Alliance of Bioversity International and CIAT, and the Forest Restoration Research Unit at Chiang Mai University (FORRU-CMU).
The first session, From Seeds to Success , brought together 95 participants from across Asia and beyond, and was heavily over-subscribed. The organisers committed to a second run , timed to suit participants in the Americas as well as Europe and Asia, and invited Infonomic to present Byline as a new addition to the programme. The response suggests that demand for practical, well-connected restoration tools is still growing.
Our contribution to the Digital Tools Webinar, organised by Kew Science and the Forest Restoration Research Unit (FORRU-CMU) at Chiang Mai University.
The tools
Each of the tools presented addresses a different stage of the restoration pipeline, from deciding what to plant through to monitoring what was planted. Taken together, they form the beginnings of a genuinely useful digital ecosystem for restoration practitioners, and we would encourage anyone working in the field to explore them.
- Diversity for Restoration (D4R) , presented by Dr Tobias Fremout (Alliance of Bioversity International and CIAT). A freely available decision-support tool that recommends tree species and seed sources matched to site conditions, restoration objectives and future climate scenarios.
- MyFarmTrees , presented by Dr Chris Kettle (Alliance of Bioversity International and CIAT). A platform that tracks restoration from seed collection, through the nursery, to trees established in the field, using QR-coded seed lots and mobile devices.
- SeedPOD, the Seed Portal for Online Data , presented by Naomi Carvey (Royal Botanic Gardens, Kew). An open-access portal, now soft launched, that standardises and shares seed collection, viability and availability data from the Millennium Seed Bank Partnership and its network of seed custodians.
- Germination Experiment Assistant (GEA), presented by Dr Jan Sala Vila (Royal Botanic Gardens, Kew). An AI-assisted system that predicts germination requirements and generates practical protocols for species with little or no published germination data.
- Wyse–Dickie Seed Storage-Behaviour Predictor , presented by Dr Kiran Dhanjal-Adams (Royal Botanic Gardens, Kew). A model that predicts whether a species' seeds are likely to tolerate drying (orthodox) or not (recalcitrant), to guide storage decisions.
Our thanks to the organisers at Kew, the Alliance and FORRU-CMU, and in particular to Dr Kate Hardwick, Dr Stephen Elliott and Dr Chris Kettle, for bringing the series together, and to all of the presenters for their generosity in discussion.
Our contribution: Byline and the FORRU library
Our part of the session was less about a single tool and more about the systems underneath them. Good restoration science only creates value when it is found and used, and a practitioner with a question rarely starts at a research unit's website; they start at a search engine.
We have worked with FORRU-CMU for several years on forru.org , which brings roughly 300 publications, together with FORRU's advice, advocacy and outreach material, into a single bilingual platform. Visitors can search inside documents as well as across titles and abstracts, filter by topic and region, and share any result set by its own URL. For FORRU's staff, an enquiry that once meant finding, scanning and emailing a paper can now be answered with a link. In Dr Stephen Elliott's words, the site offers "a sophisticated search capability, enabling users to quickly find relevant information, even within documents."
The results are measurable. In June 2026, Google Search Console recorded around 1,580 clicks from web search to forru.org, with 54 pages appearing on the first page of results, and the site received more than 3,000 unique visitors in the month, with no advertising. The reasons are unglamorous: every publication is described as research (Dublin Core, JSON-LD and Google Scholar metadata), its full text is extracted and indexed, the site is published in Thai as well as English, and every filtered view is a page that search engines can index.
forru.org runs on Byline , the open-source content management system we build at Infonomic. Byline was designed for exactly this kind of material rather than adapted to it:
- Versioning, workflow and languages in the core. Research material is corrected and superseded; even a two-person unit needs draft, review and publish; and FORRU publishes in more than one language.
- An extensible core. Publications, projects, species records and seedling keys can live in one system, and new kinds of structured data can be added without adopting another platform.
- Open source and free to use. FORRU holds its own data; the community holds the core. We charge for our work, not for the software, so the collection does not depend on us.
A better way out of the PDF
Much of our current work with FORRU is unglamorous but, we think, foundational. PDFs are a poor exchange format: headings, tables, figures and illustrations need to be extracted in a way that preserves their meaning, not as fragments of text that may not even be in the right order. FORRU's collection makes the problem concrete. Thirty years of forest restoration research spans books, book chapters, theses, dissertations, journal articles and scanned typescripts.
We are now re-extracting and re-indexing the entire FORRU collection. Each document is inspected and routed down an appropriate extraction path, combining conventional text extraction, OCR and AI-based methods, so that the result is structured, semantically correct text even for complex documents. Small institutions are benefiting here from the private sector's current race to unlock institutional knowledge; the techniques it has produced are now within reach of a research unit with a well-organised collection. Clean, well-extracted text is the minimum gate for everything that follows.
We have written up this approach in a demonstration paper, Cheap First, Escalate on Evidence: Quality-Gated Extraction for Heterogeneous Research Collections, by Anthony Bouch, Stephen Elliott and David Lipsky. The paper has been accepted for ICADL 2026 , the International Conference on Asia-Pacific Digital Libraries, hosted this year at Khon Kaen University in Thailand, and will appear in the Springer LNCS proceedings. It describes a pipeline that applies inexpensive extraction first and escalates to more costly methods only where quality checks show they are needed, using the FORRU collection as its test case.
What comes next
forru.org currently runs relevance-ranked full-text search (Solr and BM25) over the extracted collection. With the re-extracted corpus in place, the next step is semantic and hybrid search, combining the strengths of keyword search with semantic discovery and AI-supported retrieval.
The longer-term goal is a restoration assistant whose answers are verifiable: clearly attributed, and linked back to the FORRU publications they draw on. FORRU receives enquiries every week that such an assistant could help with, from "Why is biodiversity important in forest restoration?" to a landowner asking what to plant on a degraded site, which quickly becomes a series of questions about altitude, existing species, seed sources, germination and planting season. We want to be candid about where this stands. The foundation is built, but the assistant is not; and it will only ever be as good as the corpus beneath it, which is why the extraction work comes first.
We are grateful to FORRU-CMU, Kew and the Alliance of Bioversity International and CIAT for including us, and to everyone who registered and took part. If you would like to continue any of these conversations, you can find us at infonomic.io , and Byline at bylinecms.app .