Reconstructing past bumblebee foraging networks to guide landscape restoration
Bumblebees are declining across the UK, but restoring landscapes for threatened species is difficult because we often do not know which floral resources they used when their populations were more widespread. Modern observations tell us which plants bumblebees use in today’s landscapes, but they cannot reveal resources that have disappeared through more than a century of land-use change.
A remarkable record of these historical interactions survives in naturalist writing from the Victorian period onwards, including county faunas, field notes and local natural-history accounts. Much of this material has never been systematically converted into ecological data. Recent advances in artificial intelligence and large language models now make it possible to recover this information at a scale that would previously have been impractical.
This PhD will reconstruct UK bumblebee–plant interactions from the Victorian period to the present and investigate how changes in land use have reshaped bumblebee foraging networks. The student will combine three complementary sources of evidence.
First, they will adapt an existing AI-assisted evidence-synthesis pipeline to extract bumblebee species interactions, locations, dates, habitats and other ecological information from historical naturalist literature. The student will develop and validate approaches for working with historical language, changing taxonomy and uncertain records, with human verification incorporated throughout the workflow.
Second, the student will work with the Bumblebee Conservation Trust (BBCT) to analyse contemporary data, including records from BeeWalk, the national bumblebee monitoring scheme. These data will be used to characterise present-day patterns of floral-resource use across species, regions and seasons.
Third, historical and contemporary interaction records will be linked with information on landscape change, including historical Ordnance Survey maps, aerial photography and modern land-cover datasets. This will allow the student to investigate how changes in habitat composition and connectivity relate to the loss, persistence or emergence of particular plant–pollinator interactions.
The project will ask whether declining bumblebee species historically used a broader or different set of floral resources than contemporary observations suggest; whether losses of particular interactions are associated with changes in land use and habitat connectivity; and whether historical interaction networks can help identify plant assemblages and seasonal forage resources that could support threatened species today. Statistical and ecological-network models will account for differences in sampling effort and observation processes between historical and modern datasets.
The project combines ecology, conservation, artificial intelligence, historical biodiversity data, geospatial analysis and ecological modelling. The student will receive training in Python and R, large language models, reproducible evidence synthesis, research software engineering, database development, GIS, statistical modelling and ecological-network analysis.
A three-month CASE placement with the Bumblebee Conservation Trust will provide direct experience of applied conservation research. The student will work with BBCT scientists to integrate historical evidence with contemporary monitoring, with opportunities to investigate threatened species such as the Shrill Carder bee and Moss Carder bee and to translate the results into practical recommendations for habitat restoration.
The project therefore offers an opportunity to develop new computational approaches to recovering ecological knowledge from the past while addressing an urgent conservation question: can understan
The student will receive training in Python/R, large language models, reproducible evidence-synthesis pipelines, database
design and management, statistical and ecological network modelling, and interpretation of historical and citizen-science
datasets. Training will combine TREES cohort activities with specialist opportunities across SBBS, Geography and the Centre
for Probability, Statistics and Data Science. The student will use version control and open, reproducible research practices
throughout, and gain experience managing large heterogeneous datasets.
Alongside technical development, the student will gain transferable skills in interdisciplinary collaboration, scientific writing,
open-source software development, project management, stakeholder communication and translating research into
conservation practice. They will present at national and international meetings, including British Ecological Society
conferences, participate in seminars and research groups across Queen Mary, and work directly with the Bumblebee
Conservation Trust.
A combination of academic, computational and applied conservation training will prepare the student
for careers spanning ecological research, environmental data science, conservation organisations and environmental policy
