eprintid: 10205502 rev_number: 10 eprint_status: archive userid: 699 dir: disk0/10/20/55/02 datestamp: 2025-03-06 12:35:16 lastmod: 2025-03-06 12:35:16 status_changed: 2025-03-06 12:35:16 type: thesis metadata_visibility: show sword_depositor: 699 creators_name: Bevan, Peggy Alice title: Challenges to biodiversity monitoring across spatial and ecological scales ispublished: inpress divisions: UCL divisions: B02 divisions: C08 divisions: D09 divisions: F99 keywords: Biodiversity, Machine learning, Camera Traps, Biogeography, Species monitoring note: Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. abstract: In the face of a biodiversity loss crisis, quantifying the impact of human-induced pressures on ecological communities is crucial for future conservation efforts. In this thesis, I investigate methods for measuring wildlife responses to anthropogenic pressure and address some of the challenges of a global biodiversity monitoring framework.. In my first analysis, I use a meta-analytical approach to reveal the unique responses of biodiversity to land-use change across biome, realm and taxonomic group. These results suggest that monitoring across regionally separated biomes (regional biomes) could improve global monitoring efforts. In the next chapters, I use a regional biome case-study in Asian sub-tropical dry forest to show nuanced responses of species to anthropogenic pressure. In Chapter 3, I use a dataset of 150 camera traps (CTs) covering a gradient of pressure in Nepal’s terai region to show that variation in species’ threat responses is producing novel co-occurrence networks. As habitat availability reduces, some species alter their diel activity patterns to reduce interspecific overlap, with implications for long-term fitness and community resilience. Collecting nuanced biodiversity metrics like this on a global scale would require efficient data collection using minimal resources. Automated analysis tools, such as deep learning image classifiers, provide a solution for rapidly processing vast amounts of data. In Chapter 4, I use two CT data sets from Kenya and Nepal to show how manipulating aspects of a deep learning classification pipeline, such as training set size or model architecture, has a low impact on the accuracy of ecological metrics. In Chapter 5, I use an unsupervised deep learning pipeline to show that ecological patterns can be extracted from CT data with no species labelling, demonstrating the potential of automated tools to transform ecological monitoring. Through the research presented here, I have identified major gaps in global biodiversity monitoring and explored innovative solutions for addressing these gaps. date: 2025-02-28 date_type: published oa_status: green full_text_type: other thesis_class: doctoral_open thesis_award: Ph.D language: eng primo: open primo_central: open_green verified: verified_manual elements_id: 2365312 lyricists_name: Bevan, Peggy lyricists_id: PBEVA10 actors_name: Bevan, Peggy actors_id: PBEVA10 actors_role: owner full_text_status: public pagerange: 1-1 pages: 190 institution: UCL (University College London) department: Genetics, Evolution and Environment thesis_type: Doctoral citation: Bevan, Peggy Alice; (2025) Challenges to biodiversity monitoring across spatial and ecological scales. Doctoral thesis (Ph.D), UCL (University College London). Green open access document_url: https://discovery.ucl.ac.uk/id/eprint/10205502/7/Bevan_10205502_thesis_redacted.pdf