Climate change, deforestation and disease: modelling triple threats to an endemic tree in a tropical biodiversity hotspot

Tropical forests harbour extraordinary biodiversity and provide critical contributions to global carbon cycling and climate regulation, yet experience multiple, interacting abiotic and biotic threats including anthropogenic climate change, land use and land cover change, and biological invasions. Whether large, long-lived tree species can track the velocity of future climate change is unclear, as lack of data makes parameterising predictive models difficult. By largely focusing on climate change, existing projections of plant species’ future range shifts and distributions often miss the complexity and nuance of interacting stressors. This thesis aims to disentangle factors affecting the range shift capacity of Calophyllum paniculatum – a vulnerable, endemic tree to Madagascar. Calophyllum’s survival likely depends on limiting the potential spread of a wilt pathogen Leptographium calophylli, while maintaining connectivity of forest habitats crucial for endangered primate lemurs, Calophyllum’s main seed disperser. After providing background knowledge in Chapter 1-3, I simulated long-distance seed dispersal events in Chapter 4 using an individual-based model, SiMRiv, with observed lemur movement and behaviour. Chapter 5 employed ensemble species distribution models (SDMs) using future bioclimatic and deforestation predictor variables with three machine learning algorithms (Random Forest, Boosted Regression Trees, and MaxEnt) to project future distributions of Calophyllum paniculatum and Leptographium calophylli. Chapter 6 incorporated previous findings to parameterise Calophyllum population dynamics using RangeShiftR, then assessed thirty scenarios representing gradients of combined future climate change, deforestation, defaunation, and pathogen spread to 2110. These were developed through altering the underlying landscape’s habitat quality (climate change and deforestation), emigration probability (defaunation) and spatial demographic scaling (pathogen spread). The SDMs identified consistent spatial overlap of both the tree and wilt species, while mechanistic scenarios highlighted the pathogen as the most direct threat to population persistence alongside climate change and deforestation, and defaunation scenarios significantly reduced abundance and occupancy. Sustained, long-term climate exposure produced the earliest extinctions of Calophyllum populations in pathogen scenarios, highlighting the synergistic effect of interacting stressors. While contributing to collective efforts to halt the climate and biodiversity crises, this thesis demonstrates that under the irreversible threat of extinction, despite inherent uncertainties and data limitations, combined predictive model frameworks are a valuable tool in analysing areas and species undergoing multiple threats. Their use can accelerate our focus to the habitats, communities and ecosystems that need targeted policy, management and practical support.