CR2025_08 AI in the Museum – quantifying over a century of change in the geographical ranges of the birds of Madagascar

Lead Supervisor: Ken Norris, Natural History Museum

Email: k.norris@nhm.ac.uk

Co-supervisors: Deepa Senapathi, Department of Sustainable Land Development, University of Reading.

We are witnessing unprecedented loss of biodiversity due to human activities. Historical records show that wild species have disappeared from places they used to occur, and in extreme cases this process has led to their extinction. In ecological terms, their geographical ranges have contracted sometimes to the point at which they disappear altogether.

Understanding how and why geographical ranges of species are changing is critical to nature conservation. It helps us identify factors, including human activities, driving these changes, and ultimately identify and implement actions to reduce human impacts.

Natural history collections, like those held at the Natural History Museum (NHM), are a rich repository of data on where particular species were found in the past but are grossly under-used in the study geographical range changes. Data are often inaccessible because they are on specimen labels or record cards that would take many years for a person to extract.

This situation is changing rapidly due to artificial intelligence. These technologies allow us to extract data from collections quickly and accurately, opening-up research that would have been largely impossible only a few years ago.

This PhD project brings together technological advances with natural history collections to explore changes in the species ranges of one of the most iconic biodiversity hotspots on earth – Madagascar.

Madagascar, a megadiverse country has a high concentration of endemic plant and animal species (i.e., species found nowhere else on earth). Nature in Madagascar is under pressure due to the loss and degradation of natural habitats. Human population growth and climate change mean these pressures will continue, if not accelerate in future.

Madagascar is home to ~250 bird species of which 115 are completely unique to this country¹. About 15% of these species are threatened with extinction, which places Madagascar in the top 20 countries globally for threatened birds. Madagascar is also one of the world’s most vulnerable countries to climate change. By 2065, scientists predict a temperature increase of 1.1–2.6 °C, along with greater extremes between wet and dry seasons¹.

This studentship will explore how the geographical ranges of the birds of Madagascar are changing; whether this is driven by changing climate and land use; and what implications these changes have for their conservation.

The student will have an opportunity to create an incredible historical dataset on the birds of Madagascar from the collections held in NHM and the American Museum of Natural History (AMNH). The Archbold-Vernay Expedition (AVE) visited Madagascar from 1929-1931². The expedition collected ~7,000 bird specimens, mainly skin preparations, that are now housed in NHM and AMNH.

The expedition recorded dates and location information for all collected specimens. The NHM specimens are not databased, but Large Language Models (LLMs) have been trialled successfully to rapidly extract data from label images. AMNH’s AVE bird specimens are already databased, and access to their database has been granted for this project. By combining the NHM and AMNH data, we will be able to create a unique dataset of the distributions of the birds of Madagascar from ~100 years ago.

This PhD project will build on existing data and this previous work in the following ways:

• Build an image library of the labels associated with the AVE specimens held in NHM
• Refine protocols and apply LLMs to rapidly extract the data from these labels, and combine the NHM and AMNH data
• Use web-based location services (e.g. GEOLOCATE) to geo-reference all the AVE specimens from their location information
• Map locations and create species distribution models (SDMs) by combining the specimen data with historic land cover and climate data (e.g. HILDA+), to generate historical ranges for as many bird species as possible.
• Access contemporary location information for Madagascar birds from the Global Biodiversity Information Facility (GBIF) and construct a new suite of SDMs to generate contemporary ranges.
• Use these analyses to explore how and why ranges have changed over the last century and use the SDMs to forecast how ranges might change under a range of future climate scenarios.

Outputs from this work will include a unique historical dataset and several high-quality papers. We plan to feed our results into conservation actions for Madagascar’s birds through contacts in Birdlife International.

Training opportunities:

The student will spend significant time at the Natural History Museum (NHM) in Tring developing skills on the curation and management of natural history collections focusing on the bird collections of which the Madagascar specimens are part. The student will also image specimens (specifically their labels), and work with NHM staff in the AI and Innovation Unit to refine existing protocols that use Large Language Models (LLMs) to extract data from these images. In addition, there may be the possibility of a study visit to the American Museum of Natural History in New York to share data and findings.

Student profile:

The project would suit a student looking to develop inter-disciplinary research skills that span computer, data and conservation science. A background in any of these areas of science at BSc and/or MSc would be suitable. Experience of carrying out a research project at MSc level or through work experience would be desirable. The student will need to have a keen interest in applying ‘big data’ to addressing significant questions in environmental change research. No previous experience working with natural history collections is required. The student will receive full training in all the approaches relevant to the PhD project.

Please note: Due to the nature of this project and to comply with visa regulations, only Home students should apply

Co-Sponsorship details:

The project will receive co-sponsorship from Natural History Museum. This co-sponsorship will be in the form of a placement.

References:

1. Birdlife International (2021) State of Madgascar’s Birds – Indicators of Environmental Change https://www.birdlife.org/wp-content/uploads/2021/10/State_of_Madagascars_Birds_English.pdf
2. Mission Zoologique Franco-Anglo-Américaine à Madagascar (1929-1931) https://data.library.amnh.org/archives-authorities/id/amnhc_2000264