Titolo della tesi: Interfaces of spillover of bat born pathogens in human dominated environments
Background:
Climate change, land-use transformation, and anthropogenic pressures pose major threats to biodiversity, with significant impacts on ecosystem services and human health (Díaz et al., 2018; Keesing & Ostfeld, 2021). Biodiversity loss can facilitate the rise of emerging infectious diseases (EIDs), most of which are zoonotic and originate with a spillover event from a wildlife reservoir towards humans (Allen et al., 2017; Schmeller et al., 2020). Zoonotic spillover is a complex process shaped by ecological, environmental, and human factors (Keesing & Ostfeld, 2021; Plowright et al., 2021). Central to this dynamic are interfaces between primary and secondary hosts of pathogens where the spillover might be facilitated or discouraged, depending on the spatial, temporal, and ecological contact between wildlife reservoirs, and humans, often mediated by domestic animals. These interfaces can be potential, where the contact between humans and the wildlife reservoirs is possible but not yet ongoing, or realized, where such contacts occur and create a functional pathway for spillover. The transition from potential to realized interfaces is driven by environmental, ecological, and pathogen-specific factors. Focusing on a One Health approach, a potential interface should guide preventive surveillance, while a realized interface requires immediate risk assessment and biosecurity interventions (Hassell et al., 2017).
The general scope of this PhD thesis is to develop and apply an integrated framework for identifying, characterizing, and assessing spillover interfaces between bats, humans and domestic animals in a human dominated environment. By moving progressively from potential to realized interfaces, this work aims to provide actionable insights for targeted surveillance, biosecurity interventions and risk mitigation strategies in the context of One Health.
Chapter 1: Climatic clues to local distributions: a hierarchical approach to model the distribution of Miniopterus schreibersii in Italy
Species distribution models (SDMs) are a cornerstone tool in ecology, conservation, and epidemiological risk assessment, yet their predictive capacity can be limited when relying exclusively on coarse-scale climatic variables. In this chapter, I modelled the distribution of the common bent-wing bat (Miniopterus schreibersii) in Italy, a troglophile species of high conservation concern and public health interest, comparing a traditional bioclimatic SDM (bio-SDM) with a hierarchical SDM (h-SDM) that integrates fine-scale ecological predictors. Using a maximum entropy approach, the bio-SDM was calibrated on broad-scale climatic variables, while the h-SDM combined these with local-scale descriptors of roosting (distance to karstic structures, proximity to urban areas), foraging ecology (canopy cover, agricultural land proportion), and the climatic suitability derived from the bio-SDM. Model performance was evaluated using AUC, TSS, and Boyce Index, and spatial disagreement was examined via regression and residual analysis. Both models performed well, but the h-SDM showed significantly higher maxTSS values (0.81 vs. 0.42) and lower, more variable Boyce Index values (0.26 vs. 0.86), indicating differences in calibration despite similar discrimination ability, and reduced overpredictions in urban and intensively cultivated areas. Climatic suitability and distance to karstic structures emerged as key predictors in refining distribution estimates. This work supports the value of integrating hierarchical modelling frameworks to improve spatial accuracy in species distribution models, with implications for conservation planning and the targeted surveillance of pathogens in bat populations under changing climate and land-use scenarios. This Chapter has been submitted as a research paper to Ecological Informatics and is currently under its second round of reviews with a comment of major revisions due on 14/1/2026.
Chapter 2: Blueprints of Risk: modelling interfaces between Miniopterus schreibersii and humans across Italy
This chapter operationalises a potential spillover interface by quantifying where the predicted distribution of the common bent-wing bat (Miniopterus schreibersii) co-occurs with human population density across Italy. I coupled a hierarchical SDM of M. schreibersii (projected at 100 m and harmonised at 1 km2) with GPWv4 human density (1 km2) and applied a bivariate Local Moran’s I (LISA) in GeoDa using a first-order rook contiguity weights matrix, 10,000 random permutations, and p-value < 0.05 to detect significant local clusters and outliers. Global spatial association was positive (Moran’s I = 0.040), while local statistics revealed a heterogeneous mosaic. HH clusters (i.e., cells with high bat suitability embedded within densely populated areas) concentrated around foothill and peri-urban belts of Northern cities (e.g., Trento, Bergamo, Brescia, Verona, Vicenza, Genova), parts of the Cinque Terre and Arezzo, Rome and adjacent municipalities, stretches of the Adriatic coast (Abruzzo, Marche and Molise regions), the metropolitan area of Naples, multiple centres in Puglia (Bari, Brindisi, Lecce, Taranto), and urban belts in Sardinia (Cagliari, Sassari, Nuoro) and Sicily (Catania, Palermo). LL clusters dominated mountain and inland regions (Alps, Apennines, interior Liguria and Piemonte, inland Marche, Abruzzo, Molise, and Basilicata, central Sardinia, interior Sicily). Outliers separated North and Centre-South: LH ( e.g. low bat suitability amid high human density) were prevalent across the Po Valley (e.g., Milano, Torino, Bologna, Veneto plain and around Treviso, Padova, Firenze), whereas HL ( e.g. high bat suitability amid low human density) were more frequent in Tuscany, Puglia, Sardinia, Sicily, the Pre-Alps, Cilento and Lucanian Apennines, and the eastern Calabrian coast. Comparison with known urban colonies showed partial agreement (Arezzo in HH), while Mantova and Cesena fell into LH outliers, indicating possible under-representation of urban roosts. Methodologically, bivariate-LISA framework yields statistically explicit maps of potential interfaces: HH clusters mark immediate surveillance priorities, whereas HL and LH outliers may flag nascent or shifting interfaces under land-use and climate change scenarios. Key limitations include reliance on SDM suitability versus real presence of the species, potential urban sampling bias, and the absence of standardized census of domestic and companion habitats (e.g., free-roaming cats) that can mediate contact; nonetheless, the approach is replicable and actionable for One-Health prioritisation of proactive field investigations. This Chapter is currently under internal review.
Chapter 3: A multi-disciplinary approach to identify spillover interfaces of bat coronaviruses to pig farms in Italy
Bats are recognised reservoirs of diverse coronaviruses (CoVs), but little is known about the pathways enabling their spillover into livestock. In this chapter, I applied a multidisciplinary approach, combining bioacoustic surveys, landscape analysis, and molecular virology, to assess the risk of CoV transmission from bats to pigs in intensive farming systems of Northern Italy. Between 2021 and 2022, I conducted bioacoustic monitoring in 14 pig farms to quantify bat presence and species diversity, while analysing farm-level and landscape variables to identify predictors of bat activity and richness. Additionally, I investigated CoV circulation in three colonies of Pipistrellus kuhlii through active longitudinal surveillance and performed whole-genome sequencing on new and archival CoV strains detected in P. kuhlii and Hypsugo savii. The surveys identified eight bat species across farms, with P. kuhlii, P. pipistrellus, and H. savii being the most widespread and active. Farm structural features attracting insects were associated with higher bat activity, whereas the surrounding habitat had little effect. Importantly, the frequent absence of physical barriers preventing contact between bats or their droppings and pig enclosures indicated a possible risk of exposure. Focusing on the most common bat species, I detected active CoV circulation in P. kuhlii, including colonies located in pig facilities, with two distinct CoV species identified, highlighting potential for viral recombination. I detected CoVs throughout the sampling season, with amplification peaks in May and August. Phylogenetic analyses indicated that pigs could be exposed to at least eight bat CoV species in Italy, with evidence of viral sharing between P. kuhlii and H. savii. This study outlines a potential transmission route of bat CoVs to swine and identifies key risk factors, including structural features, biosecurity gaps, species composition of the bats populations, viral diversity, and seasonal shedding patterns. This Chapter has been published as a research paper in PlosOne (doi: 10.1371/journal.pone.0332117)
Conclusions
Across its three chapters, this thesis proposes an integrated, multi-scale framework for identifying, characterising, and assessing spillover interfaces, moving from broad-scale climatic suitability (Chapter 1) to spatially explicit mapping of potential contact areas (Chapter 2), and finally to the direct ecological and virological characterisation of a realized interface (Chapter 3). Together, these studies corroborates that spillover risk emerges from the interplay of species’ ecological characteristics, environmental conditions, anthropogenic structures, and biosecurity gaps. At the potential interface level, combining hierarchical species distribution modelling with spatial autocorrelation analyses allowed me to pinpoint priority areas where ecological suitability and dense human presence overlap, providing a tool for targeted surveillance. At the realized interface level, this multidisciplinary fieldwork reveals that structural features within farms, rather than surrounding landscapes, are key determinants of bats presence, while molecular surveillance highlights viral circulation and seasonal shedding patterns that may facilitate cross-species transmission. These findings underscore the importance of integrating predictive modelling, spatial epidemiology, and in situ ecological and virological monitoring to bridge the gap between theoretical risk and observed transmission pathways. By focusing on bats as a model taxon, the thesis provides methodological and conceptual advances relevant to One Health surveillance and zoonotic risk prevention, emphasising that proactive intervention requires both anticipatory mapping of emerging interfaces and detailed characterisation of existing ones.