Universidade Federal de São Carlos Centro de Ciências Biológicas e da Saúde Programa de Pós-Graduação em Ecologia e Recursos Naturais BRUNO FARIA DA CUNHA BARBOSA ADORNO FIRES IN THE ATLANTIC FOREST: SELECTIVITY TOWARDS DIFFERENT ENVIRONMENTS AND IMPACTS ON BIRD DIVERSITY São Carlos - SP 2025 BRUNO FARIA DA CUNHA BARBOSA ADORNO Fires in the Atlantic Forest: Selectivity towards different environments and impacts on bird diversity Tese apresentada ao Programa de Pós- Graduação em Ecologia e Recursos Naturais da Universidade Federal de São Carlos como parte dos requisitos necessários para a obtenção do título de Doutor em Ciências. Área de Concentração: Ecologia e Recursos Naturais. Orientador: Prof. Dr. Augusto João Piratelli São Carlos - SP 2025 Dedico este trabalho à minha família Agradecimentos Aos meus pais, Nerita e Marco, por sempre me apoiarem na carreira de biólogo e pesquisador. Aos meus irmãos, Caio e Helena, por estarem sempre presentes. Ao meu orientador, Augusto João Piratelli, por me orientar ao longo de quatro anos e tornar esta jornada mais tranquila. Também agradeço aos meus coorientadores, Erica Hasui e Milton Ribeiro. Dificilmente eu teria conseguido sem vocês. A experiência de vocês três é inspiradora, e espero um dia me tornar um profissional à altura de vocês. Ao Pedro, da Universidade de Lisboa, que me supervisionou durante meu intercâmbio. A todos os colegas de laboratório, colaboradores de artigos, ajudantes de campo e todos que, de alguma forma, estiveram envolvidos nesta jornada. Sei que não é fácil passar dias subindo e descendo morros, fugindo de abelhas e pulando cercas. Tampouco é simples passar horas e horas mexendo no R e no QGIS. Muito obrigado! A todos os laboratórios de pesquisa envolvidos: LECO, ECOFRAG, LEEC, CE3C. A todas as instituições e seus respectivos membros que deram suporte a este trabalho: UFSCar, UNIFAL, UNESP, Ulisboa e Fundação Florestal. Por fim, às agências financiadoras: CAPES (finance code 001), FAPESP e CNPq (Processo 140361/2021-9). Obrigado! Eu sou o que me cerca. Se eu não me preservar o que me cerca, eu não me preservo. José Ortega y Gasset Resumo Incêndios florestais têm se tornado um distúrbio cada vez mais frequente em florestas tropicais, impulsionados pela intensificação das atividades humanas e pelas mudanças climáticas. Nesse contexto, compreender os padrões de ocorrência do fogo e seus impactos nas comunidades biológicas é essencial para subsidiar estratégias de manejo e conservação. Nesta tese de doutorado, busquei abordar essas questões no bioma da Mata Atlântica por meio de dois capítulos. No primeiro capítulo, publicado na revista Journal of Environmental Management, investiguei como o fogo seleciona diferentes tipos de uso e cobertura da terra ao longo de um período de 35 anos (1987–2022). A análise de mais de 40.000 incêndios revelou que florestas secundárias são os ambientes mais suscetíveis ao fogo, enquanto florestas maduras apresentam a menor propensão, evidenciando sua notável resistência a incêndios. Um achado particularmente relevante foi a diminuição da seletividade do fogo em florestas secundárias ao longo do tempo. Esse resultado sugere que, à medida que o processo de sucessão ecológica avança, as florestas secundárias tornam-se progressivamente mais resistentes ao fogo. No segundo capítulo, submetido à revista Journal of Applied Ecology, investiguei os impactos do fogo nas comunidades de aves da Mata Atlântica. Especificamente, analisei como o fogo influencia a composição e a diversidade, tanto taxonômica quanto funcional, dessas comunidades, bem como os efeitos de características da paisagem e dos incêndios sobre essas respostas. Os resultados indicaram que, de forma geral, a composição e a diversidade das comunidades de aves não foram significativamente alteradas pelo fogo. No entanto, a riqueza de espécies mostrou uma associação positiva com a produtividade da vegetação, enquanto a diversidade funcional foi influenciada pelo tamanho dos incêndios. Em conjunto, os dois capítulos desta tese oferecem novas evidências sobre as interações entre o fogo, a paisagem e a biodiversidade na Mata Atlântica. Os resultados ressaltam a importância de políticas públicas e estratégias de manejo que considerem as particularidades locais, incluindo iniciativas de restauração florestal e práticas de manejo inteligente do fogo. Essas ações são cruciais para mitigar os impactos crescentes dos incêndios florestais e proteger a biodiversidade deste bioma extremamente ameaçado. Palavras-chave: aves, diversidade funcional, NDVI, NBR, pastagens, severidade do fogo, tamanho do incêndio Abstract Wildfires are becoming an increasingly common disturbance in tropical forests, driven by the intensification of human activities and climate change. In this context, understanding fire patterns and its impacts on biological communities is critical to informing effective management and conservation strategies. This PhD thesis tackles these issues within the Atlantic Forest biome through two main chapters. In the first chapter, published in the Journal of Environmental Management, I explored how fire interacts with different land-use and land-cover types over a 35-year period (1987–2022). By analyzing data from more than 40,000 fires, I found that secondary forests are the most fire-prone environments, while old-growth forests are the least susceptible, demonstrating their remarkable resilience to fire. One particularly striking result was the reduced fire selectivity in secondary forests over time, suggesting that as ecological succession progresses, these forests become increasingly resistant to fire. The second chapter, submitted to the Journal of Applied Ecology, investigates how fire affects bird communities in the Atlantic Forest. I focused on how fire influences species composition and diversity—both taxonomic and functional—while also examining how landscape and fire characteristics shape these responses. The findings showed that, overall, fire did not significantly alter the composition or diversity of bird communities. However, species richness was positively linked to vegetation productivity, and functional diversity was influenced by fire size. Together, these two chapters provide valuable insights into the complex interactions between fire, landscapes, and biodiversity in the Atlantic Forest. The results highlight the need for public policies and management strategies that are tailored to local conditions, such as forest restoration initiatives and fire-smart management practices. These actions are essential for addressing the increasing threat of wildfires and preserving the biodiversity of this critically endangered biome. Keywords: birds, functional diversity, NDVI, NBR, pastures, fire severity, fire size Sumário Introduction .................................................................................................................. 13 Fire history in Brazilian tropical forests ................................................................................. 13 Fire impacts ............................................................................................................................ 16 Drivers of fire activity ............................................................................................................. 18 Fire spread in human modified landscapes ............................................................................ 20 Fire impacts on birds .............................................................................................................. 20 Fires in the Atlantic Forest ..................................................................................................... 22 Chapters of the thesis .............................................................................................................. 24 References...................................................................................................................... 25 Chapter I: Relative fire-proneness of land cover types in the Brazilian Atlantic Forest ............................................................................................................................. 39 Abstract .................................................................................................................................. 40 Graphical Abstract .................................................................................................................... Introduction ........................................................................................................................... 43 Methods .................................................................................................................................. 47 Ecoregions ........................................................................................................................... 47 Land cover maps ................................................................................................................. 49 Fire data .............................................................................................................................. 51 Data analysis ....................................................................................................................... 51 Results .................................................................................................................................... 53 Fire-proneness across ecoregions....................................................................................... 55 Fire-proneness by ecoregion ............................................................................................... 56 Fire-proneness over time .................................................................................................... 57 Discussion ............................................................................................................................... 58 Fire-proneness in the Brazilian Atlantic Forest .................................................................. 58 Fire-proneness by ecoregion ............................................................................................... 61 Fire-proneness of the Brazilian Atlantic Forest over time .................................................. 63 Management implications ................................................................................................... 65 Conclusion .............................................................................................................................. 67 Acknowledgments ................................................................................................................. 68 References .............................................................................................................................. 68 Chapter II: Fire size and vegetation productivity shape bird diversity across burned landscapes in the Brazilian Atlantic Forest .................................................. 82 Graphical Abstract ................................................................................................................ 83 Abstract .................................................................................................................................. 84 Introduction ........................................................................................................................... 85 Methods .................................................................................................................................. 88 Study region ........................................................................................................................ 88 Study sites and landscape metrics ....................................................................................... 90 Bird sampling ...................................................................................................................... 92 Bird species richness and functional diversity .................................................................... 93 Data analysis ....................................................................................................................... 94 Results .................................................................................................................................... 96 Discussion ............................................................................................................................... 98 Patterns of similarity between unburned and burned forests .............................................. 99 Bird responses to landscape features ................................................................................ 100 Management implications ................................................................................................. 102 Conclusions .......................................................................................................................... 104 Acknowledgments ............................................................................................................... 104 References ............................................................................................................................ 105 13 Introduction The year was 2019. The news on television and the internet was always the same: Brazil was on fire. Whether in the Amazon, the Cerrado, or any other biome, the Brazilian landscapes looked identical: a gray sea with waves of smoke. Images of charred animals circulated throughout the country. Who could forget the image of a carbonized jaguar? Today, six years later, the situation remains unchanged. This scenario motivated me to work on this topic and, in some way, seek solutions to mitigate (even if modestly) this environmental issue. This doctoral thesis, focused on the Atlantic Forest biome, pursued two primary objectives: (i) to analyze fire selectivity among different land-use and land-cover types (LULCs) and (ii) to assess the impacts of fire on bird communities. In this general introduction, I examine the history of fire in tropical forests, with a particular emphasis on Brazil, discussing its major environmental impacts and the key factors driving current fire activity. I also discuss fire impacts on birds, focusing on tropical forests. Finally, I provide an overview of the chapters developed throughout this research. Fire history in Brazilian tropical forests Fires are ecological disturbances with a heterogeneous global distribution across biomes. In tropical forests, fire events were historically rare, with return intervals ranging from hundreds to thousands of years, as evidenced by charcoal records (Uhl & Kauffman, 1990). This low frequency was largely attributed to the high humidity typical of these forests, which limited both fire ignition and spread (Cochrane, 2003; Cochrane et al., 1999). Until a few decades ago, the occurrence of frequent fires in tropical forests was considered unimaginable. However, the synergistic interaction between climate change and human activities has profoundly altered fire behavior (Le Page et al., 2017). Today, 14 fire frequency has increased dramatically, with return intervals reduced to decades or even shorter periods (Cochrane, 2003). These changes not only increase the number of fires but also alter their dynamics, intensifying their extent and severity (Bowman et al., 2009). In Brazil, fires were only recognized as a significant environmental issue in the late 1990s, spurred by growing awareness of the environmental and financial damages resulting from the historical misuse and indiscriminate application of fire (Soares et al., 2009). A landmark event during this period was the great fire in the state of Roraima in 1997 (INPE, 1998), which devastated approximately 11,000 km². This incident garnered national and international attention, marking one of the first clear indicators of how the interaction between human activities, such as deforestation and slash-and-burn practices, and climate change was reshaping fire dynamics. Studies from that time had already warned of the rising incidence of fires in Brazil’s tropical forests and the severe environmental challenges this would entail (Cochrane et al., 1999). These predictions have been alarmingly validated. In 2024, Brazil experienced nearly 280,000 fire events, with a burned area of approximately 22 million hectares (INPE, 2024; Mapbiomas, 2024). These events received extensive coverage from Brazilian media outlets, which highlighted not only the alarming scale of the fires but also their devastating impacts on fauna, flora, and ecosystem services. Reports underscored widespread biodiversity loss, increased carbon emissions, and severe threats to local communities and indigenous territories (Figure 1). 15 Figure 1. News reports from 2024 highlighting forest fires in Brazil. These news articles were retrieved from Google on January 13, 2025, and include: "Brazil reports 1M wildfires over last 5 years" (Agência Brasil, https://agenciabrasil.ebc.com.br/en/meio-ambiente/noticia/2025-01/brazil-reports-1m-wildfires- over-last-5-years); "State in the Amazon had the highest number of fires in Brazil in 2024" (Revista Cenarium, https://revistacenarium.com.br/en/state-in-the-amazon-had-the-highest-number-of-fires-in- brazil-in-2024/); "Wildfires continue to rise in Brazil’s main biomes in 2024" (WWF Brasil, https://www.wwf.org.br/?90121/Wildfires-continue-to-rise-in-Brazils-main-biomes-in-2024); and "Wildfires spread and reach all regions of Brazil" (Folha de S.Paulo, https://www1.folha.uol.com.br/internacional/en/scienceandhealth/2024/09/wildfires-spread-and-reach-all- regions-of-brazil.shtml). 16 Fire impacts In Brazil, most fires occur in agricultural landscapes, where they are intentionally set for purposes such as pest control, clearing stubble, renewing pastures, or expanding agricultural frontiers (da Silva Junior et al., 2020). Other causes of fires include the release of fire balloons, acts of arson by criminals or pyromaniacs, and natural factors like lightning strikes (Carvalho et al., 2022). These anthropogenic and natural fires often escape into adjacent forests, which lack the evolutionary adaptations to tolerate or recover from such disturbances (Pivello et al., 2021). Forest fires typically manifest as low-intensity surface fires, burning through the dry litter layer on the forest floor. While such fires primarily affect smaller trees and seedlings, causing significant mortality in these groups (Barlow et al., 2002; Uhl & Kauffman, 1990), their impacts can escalate under extreme drought conditions. During these periods, surface fires can intensify into high-severity fires that kill larger, thin- barked trees, further degrading the forest structure and resilience (Uhl & Kauffman, 1990). Once a forest is burned, its vulnerability to future fires increases significantly. The fragmented canopy allows for greater solar heating, altering the microclimate and facilitating the invasion of fire-adapted grass species. This creates conditions where fires become recurrent and endemic (Cochrane et al., 1999). Such altered fire dynamics often develop into a self-reinforcing cycle: each fire event exacerbates fuel accumulation on the forest floor and intensifies the severity of subsequent fires (Cochrane, 1999). The ecological consequences of these fires are profound. They drive changes in biodiversity, often resulting in species loss or shifts in community composition (Kelly et al., 2017, 2020). Fire events also degrade soil quality (Nadporozhskaya et al., 2018), increase erosion rates (Girona-García et al., 2024), and negatively affect freshwater ecosystems, such as rivers and lakes (Vaz et al., 2014, 2021). Furthermore, fires diminish 17 water retention capacity (Sansevero et al., 2020; Schmerbeck & Fiener, 2015) and alter landscape dynamics, with cascading effects on vegetation cover, carbon storage, and ecosystem services (Moreira et al., 2009; Silva et al., 2009). Fire impacts do not rely only on the environment. The economic and public health impacts of fires in Brazil are profound, affecting various sectors and posing significant challenges to sustainable development. Economically, fires lead to substantial losses in gross domestic product (GDP), agricultural productivity, and ecosystem services, particularly in regions like the Pantanal and Amazon. For example, the Pantanal fires of 2019 and 2020 resulted in GDP declines of -0.79% in Mato Grosso do Sul and -0.98% in Mato Grosso, with the livestock and agricultural sectors being the most affected (Scur et al., 2023). Fires in the Amazon cause annual losses of approximately $39 ± 2 per hectare, escalating to $183 ± 30 in areas with recurrent fires (Strand et al., 2018). These losses are further exacerbated by the negative correlation between fires and agricultural outputs, particularly for key crops like soybeans and corn, which are vital to Brazil's economy (Inacio et al., 2022). Additionally, smallholders often face a fire-poverty trap, where reliance on fire-based practices reduces yields and perpetuates economic stagnation (Cammelli et al., 2020). The public health impacts are equally severe, as fires significantly increase air pollution through the release of particulate matter (PM2.5), which has been linked to respiratory and cardiovascular diseases. During wildfire waves, hospital admissions for respiratory illnesses rise by 23%, while circulatory diseases increase by 21% (Requia et al., 2021). Fires in the Amazon in 2019 alone were responsible for approximately 3,400 additional deaths due to PM2.5 exposure (Butt et al., 2021). These health risks are exacerbated by deforestation, which not only drives fire activity but also contributes to 18 carbon emissions and biodiversity loss, hindering long-term economic and environmental sustainability (Cammelli et al., 2020). Drivers of fire activity Fire behavior in tropical forests is primarily driven by human activities and climate dynamics. Among these factors, the expansion of agricultural frontiers plays a pivotal role in increasing fire activity. Agricultural practices often rely on fire as a cost-effective tool for land clearing and management, particularly in regions undergoing rapid agricultural expansion, such as the Amazon and Cerrado biomes in Brazil, as well as parts of Asia (Carlson et al., 2012; Morton et al., 2008). This process significantly alters landscape composition by replacing undisturbed forests with more flammable land use and land cover types (LULCs). In the Amazon and Atlantic Forest biomes, for example, pastures exhibit the highest fire ignition rates and frequently act as conduits for fire spread into adjacent forested areas, further exacerbating deforestation and forest degradation (Gutiérrez-Vélez et al., 2014). On the other hand, some evidence suggests that eucalyptus plantations may offer partial protection against fire. Plantation owners, motivated by the need to safeguard their investments, actively suppress and control fires that could damage their eucalyptus crops, thereby creating a fire buffer effect in certain areas (Guedes et al., 2020). Agricultural expansion also drives changes in landscape configuration, influencing fire behavior primarily through the processes of fragmentation and the resulting edge effects. Fragmentation refers to the breaking apart of large, contiguous forested areas into smaller, isolated patches, often separated by anthropogenic LULCs (Driscoll et al., 2021). As forests are fragmented, human activities tend to increase significantly in these areas. This increased human occupation and use intensifies pressures on fragmented 19 ecosystems, resulting in higher rates of degradation and an increased risk of forest fires (Driscoll et al., 2021). Moreover, fragmentation amplifies edge effects—environmental changes that occur at the interface between forested areas and adjacent open or modified landscapes. Forest edges are more exposed to external factors such as wind, direct sunlight, and reduced humidity, which create microclimates that differ significantly from the forest interior. These altered conditions heighten the vulnerability of forest edges to degradation, increase tree mortality, promote the accumulation of combustible materials, and make these areas more prone to fires (Driscoll et al., 2021). Notably, previous studies have shown that up to 90% of forest fires are associated with forest edges in tropical forests (Cochrane, 2001). Such changes in forest edges reduce the ability of tropical forests to recycle water, making them drier (Spracklen et al., 2012) and therefore more susceptible to fires. In general, both local- and large-scale climate conditions are closely linked to fire activity in tropical forests. Previous studies have documented that fire activity in these regions is associated with decreased precipitation and extended dry seasons (Aragão et al., 2018; Gutiérrez-Velez et al., 2014). Precipitation plays a crucial role in influencing wildfire occurrence, both directly and indirectly. The amount and distribution of rainfall are key factors in determining the start, duration, and end of the fire season or periods of heightened fire risk (Aragao et al., 2018). Climate change and more intense El Niño events further exacerbate this dynamic by increasing the intensity and duration of dry seasons, thereby creating conditions that significantly heighten fire activity (Aragão et al., 2018). These prolonged and severe dry periods reduce soil and vegetation moisture, facilitating fire ignition and spread while amplifying the scale and severity of wildfire events. 20 Fire spread in human modified landscapes Fires can spread from a local epicenter to cover extensive areas, with their propagation rate being either enhanced or hindered by the spatial arrangement of fuel across the landscape. The amount and spatial distribution of fuel are fundamental factors in explaining fire ignition and propagation. Discontinuities in fuel loads lead to variations in fire-propagation rates (Carmo et al., 2011a). Consequently, fires play different roles across the components of landscape mosaics and at various stages of land-cover change trajectories (Eva & Lambin, 2000). This is because land-cover types are closely linked to fuel characteristics (Turner et al., 1997). If all LULCs within a landscape were equally fire-prone, fires would occur randomly. However, research shows that certain LULCs are more susceptible to fire than others (Carmo et al., 2011; Moreira et al., 2009). The spatial arrangement of LULCs influences fire occurrence patterns, as the probability of both fire ignition and spread varies across these types (Moreira et al., 2001). For instance, in Mediterranean areas, shrublands, grasslands, and coniferous forests are more prone to fire than croplands and broadleaf forests (Oliveira et al., 2014). In Brazil, studies indicate that pastures account for the largest proportion of burned areas (Alencar et al., 2022). However, it remains unclear whether this is due to the extensive presence of pastures in the Brazilian landscape or their inherent high fire susceptibility. Understanding fire selectivity toward specific LULCs is critical for informing policy decisions, as LULCs, unlike other factors such as topography or weather, can be actively managed. Fire impacts on birds Fire behavior influences the distribution and abundance of species, population sizes, and the availability of critical resources such as food and shelter (Barlow & Peres, 21 2004; Chia, 2015; González et al., 2022). Additionally, fires can impact various ecological interactions, including competition and predation (Letnic et al., 2013). Consequently, it is expected that fires will affect bird communities. However, bird responses to fire can vary widely depending on the species, population characteristics, and locations (González et al., 2022). Bird responses to fires can range from negative to positive (Fontaine & Kennedy, 2012). During fire events, most birds are able to escape due to their high mobility. However, nestlings, weak fliers, and ground-dwelling species may be unable to flee, making them vulnerable to the direct effects of smoke and flames, which can lead to mortality (Mardiastuti, 2020). On the other hand, some species may respond positively to fires: raptors often circle above the flames, preying on small mammals escaping the burning areas, while insectivorous birds exploit the smoke column to catch insects (Mardiastuti, 2020). Most impacts on birds are associated with post-fire changes in vegetation. Some species are closely tied to the complex vegetation structures found in unburned areas, while others benefit from the open habitats typically created by fire (Barlow & Peres, 2004). Consequently, fires often drive shifts in species composition, replacing highly specialized and rare species with those better adapted to disturbed environments (Barlow et al., 2002; Mestre et al., 2013). These effects can persist for years, with the restructuring of bird assemblages varying by region. In the Amazon, it can take three to more than ten years for bird communities to recover after a fire (Barlow et al., 2002; Mestre et al., 2013). In contrast, tropical forests in Southeast Asia may experience a complete restructuring of bird assemblages within five years of a fire event (Adeney et al., 2006). Variation in bird recovery is closely tied to the landscape characteristics of different regions. Native forest cover, for instance, plays a vital role in shaping bird communities 22 during and after fires by providing ecological refuges and serving as sources for recolonization (Robinson et al., 2014; Watson et al., 2012). In contrast, vegetation productivity serves as a proxy for resource availability, influencing bird presence in burned landscapes (Leveau & Isla, 2021; Pettorelli et al., 2005). Additionally, fire characteristics—such as size and severity—are strongly linked to changes in resource availability, often disrupting post-fire occupancy patterns and affecting landscape connectivity (Parkins et al., 2018; Steel et al., 2018, 2022). Regional variations in these factors likely reflect differences in the time required for bird communities to recover after fire events. Fires in the Atlantic Forest The Atlantic Forest (AF) biome, one of the world's most critical biodiversity hotspots, faces severe challenges due to fire and human-induced disturbances. Historically spanning 160 million hectares across Brazil, Argentina, and Paraguay (Muylaert et al., 2018), the AF has suffered centuries of intense deforestation and land- use changes, leading to the widespread loss of old-growth forests. Today, only about 36% of its natural vegetation remains, much of it fragmented into small patches under 50 hectares, isolated, and heavily impacted by human activities (Rezende et al., 2018; Ribeiro et al., 2009; Vancine et al., 2024). These fragmented landscapes, with their exposed and drier edges, are increasingly susceptible to ignition, particularly in areas adjacent to human-modified environments (Cochrane et al., 1999). Fire poses a particularly serious threat to this already vulnerable biome. Fragmentation and fire interact in a destructive feedback loop: recurrent fires degrade forest fragments, delaying or halting regeneration and creating conditions that favor further fire events (Cochrane et al., 1999; dos Santos et al., 2019; Driscoll et al., 2021). 23 This cycle can result in large-scale ecological shifts, transforming forests into non-forest ecosystems and jeopardizing the biome's unique biodiversity and resilience (dos Santos et al., 2019; Sansevero et al., 2020). Fires also disrupt the natural transition of secondary forests towards old-growth vegetation, resetting regeneration processes and potentially pushing the biome toward a savanna-like state (dos Santos et al., 2019; Sansevero et al., 2020a). Despite these threats, conservation efforts over recent decades have led to some improvements in vegetation cover. Government policies, such as the Atlantic Forest Law (2006) and the Brazilian Forest Code (2012), along with reforestation and natural regeneration driven by agricultural abandonment and rural depopulation, have contributed to the recovery of secondary forests (Vancine et al., 2024). However, these measures have not been sufficient to address the growing role of fire in shaping the AF landscape, underscoring the urgent need for targeted strategies to mitigate fire impacts and safeguard the biome's extraordinary biodiversity. In Brazil, fire-related research has predominantly focused on the Amazon and Cerrado biomes, leaving the Atlantic Forest relatively understudied. This knowledge gap limits our understanding of how fire behaves in the AF’s highly fragmented and human- modified landscapes, which differ significantly from the more extensive and continuous forest cover of the Amazon. The degraded state of the AF likely alters fire dynamics and their ecological consequences, highlighting the importance of investigating fire behavior and its impacts in this fire-sensitive biome. Regarding avian communities, research on fire's effects in the AF is remarkably scarce. To date, only one study has examined the interaction between fire and birds in this biome, focusing on the sensitivity of understory species in burned versus unburned areas 24 (Loures-Ribeiro et al., 2011). This significant gap in knowledge underscores the urgent need for research on fire’s broader ecological consequences in the Atlantic Forest, particularly given its status as one of the planet's most biodiverse and threatened ecosystems (Myers et al., 2000; Rezende et al., 2018). Chapters of the thesis This doctoral thesis is divided into two chapters. In the first chapter, I examined how different LULCs are selected by fire across the various ecoregions of the Atlantic Forest. To achieve this, over 40,000 fires were analyzed over a 35-year period. This chapter, titled "Relative fire-proneness of land cover types in the Brazilian Atlantic Forest," has been published in the Journal of Environmental Management (https://doi.org/10.1016/j.jenvman.2025.124066). In the second chapter, I investigated how fires impact the taxonomic and functional diversity of bird communities in the Atlantic Forest. For this, we sampled bird communities in the Cantareira-Mantiqueira Ecological Corridor in São Paulo State. My goal was to assess whether there are differences in bird species composition between burned and unburned forests. Additionally, I evaluated whether diversity indices respond to variables such as forest cover, primary productivity, fire size, and fire severity. This chapter is titled ‘Fire size and vegetation productivity shape bird diversity across burned landscapes in the Brazilian Atlantic Forest’ and was submitted to the Journal of Applied Ecology. 25 References Abreu, M. C., Lyra, G. B., de Oliveira-Júnior, J. F., Souza, A., Pobočíková, I., de Souza Fraga, M., & Abreu, R. C. R. (2022). Temporal and spatial patterns of fire activity in three biomes of Brazil. Science of The Total Environment, 844, 157138. https://doi.org/10.1016/j.scitotenv.2022.157138 Adeney, J. M., Ginsberg, J. R., Russell, G. J., & Kinnaird, M. F. (2006). Effects of an ENSO-related fire on birds of a lowland tropical forest in Sumatra. Animal Conservation, 9(3), 292–301. https://doi.org/10.1111/j.1469-1795.2006.00035.x Alencar, A. A. C., Arruda, V. L. S., Silva, W. V. da, Conciani, D. E., Costa, D. P., Crusco, N., Duverger, S. G., Ferreira, N. C., Franca-Rocha, W., Hasenack, H., Martenexen, L. F. M., Piontekowski, V. J., Ribeiro, N. V., Rosa, E. R., Rosa, M. R., dos Santos, S. M. B., Shimbo, J. Z., & Vélez-Martin, E. (2022). Long-term landsat-based monthly burned area dataset for the Brazilian biomes using deep learning. Remote Sensing, 14(11), 2510. https://doi.org/10.3390/rs14112510 Aragão, L. E. O. C., Anderson, L. O., Fonseca, M. G., Rosan, T. M., Vedovato, L. B., Wagner, F. H., Silva, C. V. J., Silva Junior, C. H. L., Arai, E., Aguiar, A. P., Barlow, J., Berenguer, E., Deeter, M. N., Domingues, L. G., Gatti, L., Gloor, M., Malhi, Y., Marengo, J. A., Miller, J. B., … Saatchi, S. (2018). 21st Century drought-related fires counteract the decline of Amazon deforestation carbon emissions. Nature Communications, 9(1), 536. https://doi.org/10.1038/s41467-017- 02771-y Balch, J. K., Brando, P. M., Nepstad, D. C., Coe, M. T., Silvério, D., Massad, T. J., Davidson, E. A., Lefebvre, P., Oliveira-Santos, C., Rocha, W., Cury, R. T. S., Parsons, A., & Carvalho, K. S. (2015). The susceptibility of southeastern Amazon 26 forests to fire: insights from a large-scale burn experiment. BioScience, 65(9), 893– 905. https://doi.org/10.1093/biosci/biv106 Barlow, J., Haugaasen, T., & Peres, C. A. (2002). Effects of ground fires on understorey bird assemblages in Amazonian forests. Biological Conservation, 105(2), 157–169. https://doi.org/10.1016/S0006-3207(01)00177-X Barlow, J., & Peres, C. A. (2004). Avifaunal responses to single and avifaunal responses to single and recurrent wildfires in Amazonian forests. Ecological Applications, 14(5), 1358–1373. https://doi.org/10.1890/03-5077 Bowman, D. M. J. S., Balch, J. K., Artaxo, P., Bond, W. J., Carlson, J. M., Cochrane, M. A., D’Antonio, C. M., DeFries, R. S., Doyle, J. C., Harrison, S. P., Johnston, F. H., Keeley, J. E., Krawchuk, M. A., Kull, C. A., Marston, J. B., Moritz, M. A., Prentice, I. C., Roos, C. I., Scott, A. C., … Pyne, S. J. (2009). Fire in the earth system. In Science (Vol. 324, Issue 5926, pp. 481–484). https://doi.org/10.1126/science.1163886 Brancalion, P. H. S., Niamir, A., Broadbent, E., Crouzeilles, R., Barros, F. S. M., Almeyda Zambrano, A. M., Baccini, A., Aronson, J., Goetz, S., Reid, J. L., Strassburg, B. B. N., Wilson, S., & Chazdon, R. L. (2019). Global restoration opportunities in tropical rainforest landscapes. Science Advances, 5(7), 3223–3226. https://doi.org/10.1126/sciadv.aav3223 Brando, P. M., Balch, J. K., Nepstad, D. C., Morton, D. C., Putz, F. E., Coe, M. T., Silvério, D., Macedo, M. N., Davidson, E. A., Nóbrega, C. C., Alencar, A., & Soares-Filho, B. S. (2014). Abrupt increases in Amazonian tree mortality due to drought–fire interactions. Proceedings of the National Academy of Sciences, 111(17), 6347–6352. https://doi.org/10.1073/pnas.1305499111 Brando, P. M., Oliveria-Santos, C., Rocha, W., Cury, R., & Coe, M. T. (2016). Effects 27 of experimental fuel additions on fire intensity and severity: unexpected carbon resilience of a neotropical forest. Global Change Biology, 22(7), 2516–2525. https://doi.org/10.1111/gcb.13172 Butt, E. W., Conibear, L., Knote, C., & Spracklen, D. V. (2021). Large Air Quality and Public Health Impacts due to Amazonian Deforestation Fires in 2019. GeoHealth, 5(7). https://doi.org/10.1029/2021GH000429 Cammelli, F., Garrett, R. D., Barlow, J., & Parry, L. (2020). Fire risk perpetuates poverty and fire use among Amazonian smallholders. Global Environmental Change, 63, 102096. https://doi.org/10.1016/j.gloenvcha.2020.102096 Carlson, K. M., Curran, L. M., Ratnasari, D., Pittman, A. M., Soares-Filho, B. S., Asner, G. P., Trigg, S. N., Gaveau, D. A., Lawrence, D., & Rodrigues, H. O. (2012). Committed carbon emissions, deforestation, and community land conversion from oil palm plantation expansion in West Kalimantan, Indonesia. Proceedings of the National Academy of Sciences, 109(19), 7559–7564. https://doi.org/10.1073/pnas.1200452109 Carmo, M., Moreira, F., Casimiro, P., & Vaz, P. (2011). Land use and topography influences on wildfire occurrence in northern Portugal. Landscape and Urban Planning, 100(1–2), 169–176. https://doi.org/10.1016/j.landurbplan.2010.11.017 Carvalho, L. Z. G., Massi, K. G., Coutinho, M. P., & Magalhães, V. D. (2022). Fire effects on Atlantic Forest sites from a composition, structure and functional perspective. Brazilian Journal of Biology, 82. https://doi.org/10.1590/1519- 6984.268185 Chia, E. K. (2015). Wildfire, Landscape Heterogeneity and Fauna in Fire-prone Forests. Cochrane, M. A. (2001). Synergistic interactions between habitat fragmentation and fire 28 in evergreen tropical forests. Conservation Biology, 15(6), 1515–1521. https://doi.org/10.1046/j.1523-1739.2001.01091.x Cochrane, M. A. (2003). Fire science for rainforests. Nature, 421(6926), 913–919. https://doi.org/10.1038/nature01437 Cochrane, M. A., Alencar, A., Schulze, M. D., Souza, C. M., Nepstad, D. C., Lefebvre, P., & Davidson, E. A. (1999). Positive feedbacks in the fire dynamic of closed canopy tropical forests. Science, 284(5421), 1832–1835. https://doi.org/10.1126/science.284.5421.1832 da Silva Junior, C. A., Teodoro, P. E., Delgado, R. C., Teodoro, L. P. R., Lima, M., de Andréa Pantaleão, A., Baio, F. H. R., de Azevedo, G. B., de Oliveira Sousa Azevedo, G. T., Capristo-Silva, G. F., Arvor, D., & Facco, C. U. (2020). Persistent fire foci in all biomes undermine the Paris Agreement in Brazil. Scientific Reports, 10(1), 16246. https://doi.org/10.1038/s41598-020-72571-w de Assis Barros, L., Mendonça, B. A. F. de, Sothe, C., Fernandes Filho, E. I., & Elkin, C. (2022). Fire in the Atlantic Rainforest: an analysis of 20 years of fire foci distribution and their social-ecological drivers. Geocarto International, 37(16), 4737–4761. https://doi.org/10.1080/10106049.2021.1899301 de Santana, R. O., Delgado, R. C., & Schiavetti, A. (2020). The past, present and future of vegetation in the Central Atlantic Forest Corridor, Brazil. Remote Sensing Applications: Society and Environment, 20, 100357. https://doi.org/10.1016/j.rsase.2020.100357 dos Santos, C. O. dos, Mesquita, V. V., Parente, L. L., Pinto, A. de S., & Ferreira, L. G. (2022). Assessing the wall-to-wall spatial and qualitative dynamics of the brazilian pasturelands 2010–2018, based on the analysis of the Landsat data archive. Remote Sensing, 14(4), 1024. https://doi.org/10.3390/rs14041024 29 dos Santos, J. F. C., Gleriani, J. M., Velloso, S. G. S., de Souza, G. S. A., do Amaral, C. H., Torres, F. T. P., Medeiros, N. D. G., & dos Reis, M. (2019). Wildfires as a major challenge for natural regeneration in Atlantic Forest. Science of The Total Environment, 650, 809–821. https://doi.org/10.1016/j.scitotenv.2018.09.016 Driscoll, D. A., Armenteras, D., Bennett, A. F., Brotons, L., Clarke, M. F., Doherty, T. S., Haslem, A., Kelly, L. T., Sato, C. F., Sitters, H., Aquilué, N., Bell, K., Chadid, M., Duane, A., Meza‐Elizalde, M. C., Giljohann, K. M., González, T. M., Jambhekar, R., Lazzari, J., … Wevill, T. (2021). How fire interacts with habitat loss and fragmentation. Biological Reviews, 96(3), 976–998. https://doi.org/10.1111/brv.12687 Duguy, B., Alloza, J. A., Röder, A., Vallejo, R., & Pastor, F. (2007). Modelling the effects of landscape fuel treatments on fire growth and behaviour in a Mediterranean landscape (eastern Spain). International Journal of Wildland Fire, 16(5), 619. https://doi.org/10.1071/WF06101 Eva, H., & Lambin, E. F. (2000). Fires and land‐cover change in the tropics:a remote sensing analysis at the landscape scale. Journal of Biogeography, 27(3), 765–776. https://doi.org/10.1046/j.1365-2699.2000.00441.x Fontaine, J. B., & Kennedy, P. L. (2012). Meta-analysis of avian and small-mammal response to fire severity and fire surrogate treatments in U.S. Fire-prone forests. Ecological Applications, 22(5), 1547–1561. https://doi.org/10.1890/12-0009.1 Girona-García, A., Vieira, D., Doerr, S., Panagos, P., & Santín, C. (2024). Into the unknown: The role of post-fire soil erosion in the carbon cycle. Global Change Biology, 30(6). https://doi.org/10.1111/gcb.17354 González, T. M., González-Trujillo, J. D., Muñoz, A., & Armenteras, D. (2022). Effects of fire history on animal communities: a systematic review. Ecological Processes, 30 11(1), 11. https://doi.org/10.1186/s13717-021-00357-7 Guedes, B. J., Massi, K. G., Evers, C., & Nielsen-Pincus, M. (2020). Vulnerability of small forest patches to fire in the Paraiba do Sul River Valley, southeast Brazil: implications for restoration of the Atlantic Forest biome. Forest Ecology and Management, 465, 118095. https://doi.org/10.1016/j.foreco.2020.118095 Gutiérrez-Vélez, V. H., Uriarte, M., DeFries, R., Pinedo-Vásquez, M., Fernandes, K., Ceccato, P., Baethgen, W., & Padoch, C. (2014). Land cover change interacts with drought severity to change fire regimes in Western Amazonia. Ecological Applications, 24(6), 1323–1340. https://doi.org/10.1890/13-2101.1 Gutiérrez-Velez, V. H., Uriarte, M., Defries, R., Pinedo-Vasquez, M., Fernandes, K., Ceccato, P., Baethgen, W., & Padoch, C. (2014). Land cover change interacts with drought severity to change fire regimes in Western Amazonia. Ecological Applications, 24(6), 1323–1340. https://doi.org/10.1890/13-2101.1 Hasselquist, N. J., Allen, M. F., & Santiago, L. S. (2010). Water relations of evergreen and drought-deciduous trees along a seasonally dry tropical forest chronosequence. Oecologia, 164(4), 881–890. https://doi.org/10.1007/s00442-010-1725-y Herrmann, P. B., Nascimento, V. F., Freitas, M. W. D. de, & Ometto, J. P. (2023). Fire in highland grasslands in the Atlantic Forest Biome, a burned areas time series analysis and its correlation with the legislation. Frontiers in Remote Sensing, 4. https://doi.org/10.3389/frsen.2023.1099430 Inacio, C. M. C., Nogueira, V. V., Lopes, A. M., & David, S. A. (2022). Measuring Cross-Correlations, Contagion and Long-Range Behavior between Fires in Brazil and Some Time Series Related to Its Economic Growth. Fire, 5(5), 148. https://doi.org/10.3390/fire5050148 Instituto Nacional de Pesquisas Espaciais (INPE). (1998). Estimativa da área de 31 cobertura florestal afetada pelo incêndio em Roraima, utilizando dados de multi- sensores. INPE. Kelly, L. T., Brotons, L., & McCarthy, M. A. (2017). Putting pyrodiversity to work for animal conservation. Conservation Biology, 31(4), 952–955. https://doi.org/10.1111/cobi.12861 Kelly, L. T., Giljohann, K. M., Duane, A., Aquilué, N., Archibald, S., Batllori, E., Bennett, A. F., Buckland, S. T., Canelles, Q., Clarke, M. F., Fortin, M. J., Hermoso, V., Herrando, S., Keane, R. E., Lake, F. K., McCarthy, M. A., Morán- Ordóñez, A., Parr, C. L., Pausas, J. G., … Brotons, L. (2020). Fire and biodiversity in the Anthropocene. In Science (Vol. 370, Issue 6519). American Association for the Advancement of Science. https://doi.org/10.1126/science.abb0355 Kyereh, B., Ninnoni, R., & Agyeman, V. (2007). Degraded forests are more susceptible to forest fires: Some possible ecological explanations. Journal of Science and Technology (Ghana), 26(2). https://doi.org/10.4314/just.v26i2.32987 Le Page, Y., Morton, D., Hartin, C., Bond-Lamberty, B., Pereira, J. M. C., Hurtt, G., & Asrar, G. (2017). Synergy between land use and climate change increases future fire risk in Amazon forests. Earth System Dynamics, 8(4), 1237–1246. https://doi.org/10.5194/esd-8-1237-2017 Letnic, M., Tischler, M., & Gordon, C. (2013). Desert small mammal responses to wildfire and predation in the aftermath of a La Nińa driven resource pulse. Austral Ecology, 38(7), 841–849. https://doi.org/10.1111/aec.12063 Leveau, L. M., & Isla, F. I. (2021). Predicting bird species presence in urban areas with NDVI: An assessment within and between cities. Urban Forestry and Urban Greening, 63. https://doi.org/10.1016/j.ufug.2021.127199 Loures-Ribeiro, A., Manhães, M. A., & Dias, M. M. (2011). Sensitivity of understorey 32 bird species in two different successional stages of the lowland Atlantic Forest, Brazil. Annals of the Brazilian Academy of Sciences. www.scielo.br/aabc Mardiastuti, A. (2020). Response and impact of fire on bird community in the tropical rainforest: a review. IOP Conference Series: Earth and Environmental Science, 504(1), 012001. https://doi.org/10.1088/1755-1315/504/1/012001 Marengo, J. A., Souza, C. M., Thonicke, K., Burton, C., Halladay, K., Betts, R. A., Alves, L. M., & Soares, W. R. (2018). Changes in Climate and Land Use Over the Amazon Region: Current and Future Variability and Trends. In Frontiers in Earth Science (Vol. 6). Frontiers Media S.A. https://doi.org/10.3389/feart.2018.00228 Mestre, L. A. M., Cochrane, M. A., & Barlow, J. (2013). Long-term changes in bird communities after wildfires in the central Brazilian Amazon. Biotropica, 45(4), 480–488. https://doi.org/10.1111/btp.12026 Mirra, I. M., Oliveira, T. M., Barros, A. M. G., & Fernandes, P. M. (2017). Fuel dynamics following fire hazard reduction treatments in blue gum (Eucalyptus globulus) plantations in Portugal. Forest Ecology and Management, 398, 185–195. https://doi.org/10.1016/j.foreco.2017.05.016 Moreira de Araújo, F., Ferreira, L. G., & Arantes, A. E. (2012). Distribution Patterns of Burned Areas in the Brazilian Biomes: An Analysis Based on Satellite Data for the 2002–2010 Period. Remote Sensing, 4(7), 1929–1946. https://doi.org/10.3390/rs4071929 Moreira, F., Rego, F. C., & Ferreira, P. G. (2001). Temporal (1958-1995) pattern of change in a cultural landscape of northwestern Portugal: implications for fire occurrence. In Landscape Ecology (Vol. 16). https://doi.org/10.1023/A:1013130528470 Moreira, F., Vaz, P., Catry, F., & Silva, J. S. (2009). Regional variations in wildfire 33 susceptibility of land-cover types in Portugal: implications for landscape management to minimize fire hazard. International Journal of Wildland Fire, 18(5), 563. https://doi.org/10.1071/WF07098 Morton, D. C., Defries, R. S., Randerson, J. T., Giglio, L., Schroeder, W., & Van Der Werf, G. R. (2008). Agricultural intensification increases deforestation fire activity in Amazonia. Global Change Biology, 14(10), 2262–2275. https://doi.org/10.1111/j.1365-2486.2008.01652.x Muylaert, R. L., Vancine, M. H., Bernardo, R., Oshima, J. E. F., Sobral-Souza, T., Tonetti, V. R., Niebuhr, B. B., & Ribeiro, M. C. (2018). A note on the territorial limits of the Atlantic Forest. Oecologia Australis, 22(3), 302–311. https://doi.org/10.4257/oeco.2018.2203.09 Myers, N., Mittermeier, R. A., Mittermeier, C. G., da Fonseca, G. A. B., & Kent, J. (2000). Biodiversity hotspots for conservation priorities. Nature, 403(6772), 853– 858. https://doi.org/10.1038/35002501 Nadporozhskaya, M. A., Chertov, O. G., Bykhovets, S. S., Shaw, C. H., Maksimova, E. Y., & Abakumov, E. V. (2018). Recurring surface fires cause soil degradation of forest land: A simulation experiment with the EFIMOD model. Land Degradation and Development, 29(7), 2222–2232. https://doi.org/10.1002/ldr.3021 Nunes, M. C. S., Vasconcelos, M. J., Pereira, J. M. C., Dasgupta, N., Alldredge, R. J., & Rego, F. C. (2005). Land cover type and fire in Portugal: do fires burn land cover selectively? Landscape Ecology, 20(6), 661–673. https://doi.org/10.1007/s10980- 005-0070-8 Oliveira, S., Moreira, F., Boca, R., San-Miguel-Ayanz, J., & Pereira, J. M. C. (2014). Assessment of fire selectivity in relation to land cover and topography: a comparison between Southern European countries. International Journal of 34 Wildland Fire, 23(5), 620. https://doi.org/10.1071/WF12053 Parkins, K., York, A., & Di Stefano, J. (2018). Edge effects in fire‐prone landscapes: Ecological importance and implications for fauna. Ecology and Evolution, 8(11), 5937–5948. https://doi.org/10.1002/ece3.4076 Pettorelli, N., Vik, J. O., Mysterud, A., Gaillard, J. M., Tucker, C. J., & Stenseth, N. C. (2005). Using the satellite-derived NDVI to assess ecological responses to environmental change. In Trends in Ecology and Evolution (Vol. 20, Issue 9, pp. 503–510). https://doi.org/10.1016/j.tree.2005.05.011 Piffer, P. R., Rosa, M. R., Tambosi, L. R., Metzger, J. P., & Uriarte, M. (2022). Turnover rates of regenerated forests challenge restoration efforts in the Brazilian Atlantic forest. Environmental Research Letters, 17(4), 045009. https://doi.org/10.1088/1748-9326/ac5ae1 Pivello, V. R., Vieira, I., Christianini, A. V., Ribeiro, D. B., da Silva Menezes, L., Berlinck, C. N., Melo, F. P. L., Marengo, J. A., Tornquist, C. G., Tomas, W. M., & Overbeck, G. E. (2021). Understanding Brazil’s catastrophic fires: Causes, consequences and policy needed to prevent future tragedies. Perspectives in Ecology and Conservation, 19(3), 233–255. https://doi.org/10.1016/j.pecon.2021.06.005 Requia, W. J., Amini, H., Mukherjee, R., Gold, D. R., & Schwartz, J. D. (2021). Health impacts of wildfire-related air pollution in Brazil: a nationwide study of more than 2 million hospital admissions between 2008 and 2018. Nature Communications, 12(1). https://doi.org/10.1038/s41467-021-26822-7 Rezende, C. L., Scarano, F. R., Assad, E. D., Joly, C. A., Metzger, J. P., Strassburg, B. B. N., Tabarelli, M., Fonseca, G. A., & Mittermeier, R. A. (2018). From hotspot to hopespot: An opportunity for the Brazilian Atlantic Forest. Perspectives in Ecology 35 and Conservation, 16(4), 208–214. https://doi.org/10.1016/j.pecon.2018.10.002 Ribeiro, M. C., Metzger, J. P., Martensen, A. C., Ponzoni, F. J., & Hirota, M. M. (2009). The Brazilian Atlantic Forest: How much is left, and how is the remaining forest distributed? Implications for conservation. Biological Conservation, 142(6), 1141–1153. https://doi.org/10.1016/j.biocon.2009.02.021 Robinson, N. M., Leonard, S. W. J., Bennett, A. F., & Clarke, M. F. (2014). Refuges for birds in fire-prone landscapes: The influence of fire severity and fire history on the distribution of forest birds. Forest Ecology and Management, 318, 110–121. https://doi.org/10.1016/j.foreco.2014.01.008 Sansevero, J. B. B., Garbin, M. L., Sánchez-Tapia, A., Valladares, F., & Scarano, F. R. (2020). Fire drives abandoned pastures to a savanna-like state in the Brazilian Atlantic Forest. Perspectives in Ecology and Conservation, 18(1), 31–36. https://doi.org/10.1016/j.pecon.2019.12.004 Schmerbeck, J., & Fiener, P. (2015). Wildfires, Ecosystem Services, and Biodiversity in Tropical Dry Forest in India. Environmental Management, 56(2), 355–372. https://doi.org/10.1007/s00267-015-0502-4 Scur, M. C., Centurião, D., Berlinck, C. N., Batista, E. K. L., Libonati, R., Rodrigues, J., Nunes, A. V., Garcia, L. C., Fernandes, G. W., Damasceno-Junior, G. A., Pereira, A. de M. M., Anderson, L., Ochoa-Quintero, J. M., Oliveira, M. da R., Ribeiro, D. B., & Roque, F. O. (2023). Economic Losses and Cross Border Effects Caused by Pantanal Catastrophic Wildfires. https://doi.org/10.2139/ssrn.4601097 Silva, J. S., Moreira, F., Vaz, P., Catry, F., & Godinho‐Ferreira, P. (2009). Assessing the relative fire proneness of different forest types in Portugal. Plant Biosystems - An International Journal Dealing with All Aspects of Plant Biology, 143(3), 597– 608. https://doi.org/10.1080/11263500903233250 36 Silva Junior, C. H. L., Heinrich, V. H. A., Freire, A. T. G., Broggio, I. S., Rosan, T. M., Doblas, J., Anderson, L. O., Rousseau, G. X., Shimabukuro, Y. E., Silva, C. A., House, J. I., & Aragão, L. E. O. C. (2020). Benchmark maps of 33 years of secondary forest age for Brazil. Scientific Data, 7(1), 269. https://doi.org/10.1038/s41597-020-00600-4 Singh, M., & Huang, Z. (2022). Analysis of forest fire dynamics, distribution and main drivers in the Atlantic Forest. Sustainability, 14(2), 992. https://doi.org/10.3390/su14020992 Soares, R. V., Batista, A. C., & Nunes, J. R. S. (2009). Incêndios florestais no Brasil: O estado da arte. UFPR. Sorrensen, C. L. (2000). Linking smallholder land use and fire activity: examining biomass burning in the Brazilian Lower Amazon. Forest Ecology and Management, 128(1–2), 11–25. https://doi.org/10.1016/S0378-1127(99)00283-2 Souza, C. M., Z. Shimbo, J., Rosa, M. R., Parente, L. L., A. Alencar, A., Rudorff, B. F. T., Hasenack, H., Matsumoto, M., G. Ferreira, L., Souza-Filho, P. W. M., de Oliveira, S. W., Rocha, W. F., Fonseca, A. V., Marques, C. B., Diniz, C. G., Costa, D., Monteiro, D., Rosa, E. R., Vélez-Martin, E., … Azevedo, T. (2020). Reconstructing Three Decades of Land Use and Land Cover Changes in Brazilian Biomes with Landsat Archive and Earth Engine. Remote Sensing, 12(17), 2735. https://doi.org/10.3390/rs12172735 Spracklen, D. V., Arnold, S. R., & Taylor, C. M. (2012). Observations of increased tropical rainfall preceded by air passage over forests. Nature, 489(7415), 282–285. https://doi.org/10.1038/nature11390 Steel, Z. L., Fogg, A. M., Burnett, R., Roberts, L. J., & Safford, H. D. (2022). When bigger isn’t better—Implications of large high‐severity wildfire patches for avian 37 diversity and community composition. Diversity and Distributions, 28(3), 439– 453. https://doi.org/10.1111/ddi.13281 Steel, Z. L., Koontz, M. J., & Safford, H. D. (2018). The changing landscape of wildfire: burn pattern trends and implications for California’s yellow pine and mixed conifer forests. Landscape Ecology, 33(7), 1159–1176. https://doi.org/10.1007/s10980-018-0665-5 Strand, J., Soares-Filho, B., Costa, M. H., Oliveira, U., Ribeiro, S. C., Pires, G. F., Oliveira, A., Rajão, R., May, P., van der Hoff, R., Siikamäki, J., da Motta, R. S., & Toman, M. (2018). Spatially explicit valuation of the Brazilian Amazon Forest’s Ecosystem Services. Nature Sustainability, 1(11), 657–664. https://doi.org/10.1038/s41893-018-0175-0 Turner, M. G., Romme, W. H., Gardner, R. H., & Hargrove, W. W. (1997). Effects of Fire Size and Pattern on Early Succession in Yellowstone National Park. Ecological Monographs, 67(4), 411. https://doi.org/10.2307/2963464 Uhl, C., & Kauffman, J. B. (1990). Deforestation, Fire Susceptibility, and Potential Tree Responses to Fire in the Eastern Amazon. Ecology, 71(2), 437–449. https://doi.org/10.2307/1940299 Vancine, M. H., Muylaert, R. L., Niebuhr, B. B., Oshima, J. E. de F., Tonetti, V., Bernardo, R., De Angelo, C., Rosa, M. R., Grohmann, C. H., & Ribeiro, M. C. (2024). The Atlantic Forest of South America: Spatiotemporal dynamics of the vegetation and implications for conservation. Biological Conservation, 291, 110499. https://doi.org/10.1016/j.biocon.2024.110499 Vaz, P. G., Dias, S., Pinto, P., Merten, E. C., Robinson, C. T., Warren, D. R., & Rego, F. C. (2014). Effects of burn status and conditioning on colonization of wood by stream macroinvertebrates. Freshwater Science, 33(3), 832–846. 38 https://doi.org/10.1086/676657 Vaz, P. G., Merten, E. C., Robinson, C. T., & Pinto, P. (2021). Severely burned wood from wildfires has low functional potential in streams. Journal of Applied Ecology, 58(6), 1346–1356. https://doi.org/10.1111/1365-2664.13872 Watson, S. J., Taylor, R. S., Nimmo, D. G., Kelly, L. T., Clarke, M. F., & Bennett, A. F. (2012). The influence of unburnt patches and distance from refuges on post-fire bird communities. Animal Conservation, 15(5), 499–507. https://doi.org/10.1111/j.1469-1795.2012.00542.x 39 Universidade Federal de São Carlos Centro de Ciências Biológicas e da Saúde Programa de Pós-Graduação em Ecologia e Recursos Naturais BRUNO FARIA DA CUNHA BARBOSA ADORNO Chapter I: Relative fire-proneness of land cover types in the Brazilian Atlantic Forest Bruno F.C.B. Adorno, Augusto J. Piratelli, Erica Hasui, Milton C. Ribeiro, Pedro G. Vaz Manuscript published at the Journal of Environmental Management (ISSN: 1095-8630) – available at https://www.sciencedirect.com/science/article/pii/S0301479725000428 São Carlos/SP 2025 40 Graphical abstract 41 Abstract Fires are increasingly affecting tropical biomes, where landscape-fire interactions remain understudied. We investigate the fire-proneness—the likelihood of a land use or land cover (LULC) type burning more or less than expected based on availability—in the Brazilian Atlantic Forest (AF). This biodiversity hotspot is increasingly affected by fires due to human activities and climate change. Using a selection ratio-based approach, we analyzed fire-LULC interactions in 40,869 fires over a 35-year period (1987-2022) across various ecoregions in the AF. Our findings revealed that secondary forests, forest areas that have regrown after major disturbances, burned 61% more than expected by chance, whereas old-growth forests, native forests that have developed over very long periods, burned 57% less than expected, highlighting a nearly inverse relationship in their fire- proneness. Interestingly, our data indicate that pastures in the AF are less prone to fire than expected, despite being considered among the land uses that burn the most in Brazil. Other LULCs showed variable fire-proneness, with some differences between ecoregions. Over time, the fire-proneness of secondary forests decreased, likely due to forest aging and changes in land management practices. We emphasize the necessity for tailored fire management strategies that address the unique vulnerabilities of secondary forests, particularly in the context of ongoing restoration efforts aimed at increasing native forests. Effective measures, including the implementation of 'fire-smart management' practices and enhancing the perceived value of secondary forests among local communities, are crucial for mitigating fire risks. Integrating these strategies with incentive-based approaches can bolster fire prevention, ensuring the long-term success of restoration programs. Our study provides a framework for understanding fire-landscape dynamics in tropical forests and offers actionable insights for practitioners working to safeguard these biomes from the escalating threat of wildfires. 42 KEYWORDS: Wildfire; Fire-proneness; Land cover; Secondary forests; Old-growth forests; Fire management; Atlantic Forest 43 Introduction Tropical forests are fire-sensitive biomes, with species lacking evolutionary adaptations to frequent fires and other aspects of fire regimes (Hoffmann et al., 2012; Pivello et al., 2021). Historically, these biomes experienced fires every few hundred years (Bush et al., 2008), but climate change and human activities have altered the frequency and nature of these events (Cochrane, 2003; Fonseca et al., 2019; Le Page et al., 2017). Fires now occur every decade or even annually, often escalating in size and severity (Cochrane, 2003; Flannigan et al., 2009; Uhl & Kauffman, 1990). Fires in tropical forests typically occur as low-severity surface fires, which consume the dry litter layer and result in the death of small trees and seedlings (Barlow et al., 2002; Staver et al., 2020; Uhl & Kauffman, 1990). These surface fires can escalate to high-severity fires, killing most large trees with thin bark. The resulting canopy openings increase sunlight penetration, which dries out the forest and creates conditions favorable to subsequent fires (Sansevero et al., 2020; Pivello et al., 2021). Hence, fires are associated with biodiversity changes (Kelly et al., 2017; Kelly et al, 2020), soil degradation (Nadporozhskaya et al., 2018), erosion (Girona-García et al., 2024), impacts on freshwater ecosystems (Vaz et al., 2014, 2021), diminished water retention (Sansevero et al., 2017; Schmerbeck & Fiener, 2015), and landscape dynamics alterations (Silva et al., 2011; Moreira et al., 2011). Fires are particularly worrying in the Atlantic Forest (AF) biome, one of the world’s most important biodiversity hotspots, which has high endemism (Myers et al., 2000). Originally, the AF covered 160 million hectares across coastal and inland portions of Brazil, Argentina, and Paraguay (Muylaert et al., 2018). This biome is the most degraded and deforested Brazilian biome, having suffered from intense land use and land cover (LULC) changes in recent centuries (Rezende et al., 2018). Most old-growth forests— native forests that have developed over long periods without significant disturbance— 44 were cleared to make way for human activities. Today, the AF retains only about 36% of its extent of natural vegetation. Most of this vegetation is highly fragmented, with 97% of forest patches being under 50 ha, isolated, and impacted by various human-induced disturbances (Ribeiro et al., 2009; Vancine et al., 2024). Despite these challenges, recent decades have seen improvements in vegetation cover due to conservation efforts led by the government and environmental organizations, aimed at reducing deforestation rates (Piffer et al., 2022). Also, the abandonment of agriculture and the depopulation of rural areas contributed to the resurgence of natural vegetation, particularly in the form of secondary forests—areas of forest that have regrown after major disturbances such as deforestation and fires, whose ecological value and recovery potential depend on the time since disturbance and the surrounding landscape context (Rezende et al., 2018; Souza et al., 2020; Vancine et al., 2024). Changes in Brazilian legislation, such as the Atlantic Forest Law in 2006 and the Brazilian Forest Code in 2012, are also expected to influence the fire-landscape dynamic over the years. However, despite strict regulations, fire is increasingly shaping the AF landscape (dos Santos et al., 2019), namely by resetting the gradual transitions from secondary forests to vegetation resembling old-growth forests. Fires disrupt regeneration processes and can shift the biome towards a savanna-like state (dos Santos et al., 2019; Sansevero et al., 2020). Fires do not occur randomly across the LULC types of a given landscape; instead, their occurrence and spread are determined by a complex interplay of human actions and natural phenomena (Brando et al., 2016; Fonseca et al., 2019; Le Page et al., 2017; Pivello, 2011). In tropical forests, most fires are caused by humans as a result of using fire to clear or expand agricultural areas (da Silva Junior et al., 2020; Gois et al., 2020; Oliveira-Júnior et al., 2020). Fire spread through the landscape is influenced by wind, topography, vegetation structure, moisture content of the vegetation and soil, amount and 45 composition of surface fuel load (Carmo et al., 2011; Moreira et al., 2009; Oliveira et al., 2014; Silva et al., 2009), and factors impacting firefighting efforts (de Assis Barros et al., 2022; Gutiérrez-Velez et al., 2014; Pivello et al., 2021). These factors interact with different LULC types, making some LULC types more susceptible to fire spread than others (Abreu et al., 2022; Fonseca et al., 2019; Gutiérrez-Vélez et al., 2014). We refer to this susceptibility as fire-proneness, reflecting the likelihood of a LULC type burning more or less frequently than expected based on its availability in the landscape. LULCs that are more fire-prone are likely to be disproportionately affected, beyond what would be expected by chance alone (Moreira et al., 2001). In Brazilian tropical forests, fires disproportionately affect LULC types, with pastures and croplands being the most likely to burn (Abreu et al., 2022; Cano‐Crespo et al., 2015; de Assis Barros et al., 2022; de Santana et al., 2020; Freitas et al., 2020; Gutiérrez-Vélez et al., 2014; Herrmann et al., 2023; Uriarte et al., 2016). Conversely, a negative correlation is expected between fire occurrence and spread and the cover of old- growth forests (de Assis Barros et al., 2022; Fonseca et al., 2019; Singh & Huang, 2022). This low fire-proneness among old-growth forests can be attributed to their remarkable retention of transpired moisture, which effectively reduces susceptibility to fire (Cochrane et al., 1999; Uhl et al., 1988). Unfortunately, ongoing transitions from old- growth forests to other types of LULC can often result in increased fire-proneness (de Assis Barros et al., 2022; Eva & Lambin, 2000; Fonseca et al., 2019; Singh & Huang, 2022). Particularly, changes from old-growth forests to secondary forests likely increase fire-proneness, as secondary forests are more vulnerable due to their differing species composition, lower moisture content, and elevated radiation and temperature conditions (Parsons et al., 2015; Fonseca et al., 2019; Gutiérrez-Vélez et al., 2014). Secondary forests in the AF typically emerge in degraded, abandoned pasturelands unsuitable for 46 agriculture and livestock, and may also regenerate on the edges of small forest fragments, thus increasing the risk of fire (Guedes et al., 2020). As secondary forests mature and develop characteristics typical of old-growth forests, their fire-proneness is expected to decrease (Lebrija-Trejos et al., 2011; Ray et al., 2010). Despite the extensive research on landscape-wildfire dynamics, critical gaps remain in our understanding of how fire-proneness varies among different LULC within Tropical Forests. In Brazil, the bulk of fire research has been conducted in the Amazon Forest, leading to a disproportionate focus that overlooks the distinct dynamics within the AF (dos Santos et al., 2019; Sansevero et al., 2020). Furthermore, no previous studies have comprehensively assessed the fire-proneness of different LULCs in Brazilian Tropical forests (see Carmo et al., 2011; Moreira et al., 2009; Oliveira et al., 2014). Instead, research has typically focused on examining the distribution of burned areas and the number of fire foci (Abreu et al., 2022; de Assis Barros et al., 2022; Gutiérrez-Vélez et al., 2014). Moreover, existing studies often overlook the heterogeneity of ecoregions in the AF, their cultural and socio-economic differences, as well as the interannual variability and drivers of fire activity, which likely impact fire dynamics. The ecoregions, including evergreen (ombrophilous) and seasonal forests, as well as transitional ecotone areas, are shaped by distinct temperature and rainfall patterns (Joly et al., 2014). Their distribution and fire proneness are also influenced by the AF's extensive longitudinal, latitudinal, and altitudinal variations (Marques et al., 2021). A lack of long-term studies further limits understanding of fire-LULC dynamics over time. In this study, we used a selection ratio-based approach (e.g., Moreira et al., 2009) to analyze relative fire-LULC interactions across the ecoregions of the AF, examining data at five-year intervals over a 35-year period. By applying this established approach to tropical biomes, our study offers a novel assessment of fire-landscape relationships in this 47 distinct region. We tested the following hypotheses. 1) Old-growth forests have low fire- proneness, irrespective of ecoregion or year analyzed. 2) Secondary forests exhibit high fire-proneness in all ecoregions compared to old-growth forests. 3) Secondary forests exhibit a decreasing trend in fire-proneness over the years as the average age of their patches increases and forests mature, resulting in more canopy closure and greater surface moisture. 4) Croplands, pastures, and other human-impacted LULCs exhibit variable fire- proneness dependent on ecoregion and year Methods We employed geospatial analyses to examine relative fire-proneness in different LULC types across the Brazilian AF ecoregions. Our workflow involved defining the ecoregions, obtaining and processing LULC and fire data, and estimating selection ratios to assess fire-proneness in each region. Ecoregions To assess how ecoregions affect fire-LULC interactions in the Brazilian AF, we analyzed the five regions that cover over 90% of the biome and record the most fires (Fig. 1). These ecoregions are defined by floristic compositions of typical genera and characteristic biological forms that recur within the same climate, potentially occurring on terrains with varied lithology but well-defined relief (IBGE, 2012). We utilized vector files from the Database of Environmental Information (IBGE; https://bdiaweb.ibge.gov.br), which employs the Brazilian Vegetation Classification to map and hierarchically organize the land cover across regions. The mapping methods are detailed in the IBGE’s Technical Manual of Brazilian Vegetation (IBGE, 2012). The ecoregions are described as follows: 1) Semideciduous seasonal forest (SSF): Covering 37% of the biome, this largest 48 ecoregion in the AF is marked by two distinct seasons—rainy summers and dry winters. During the dry season, up to 50% of tree species undergo leaf shedding. 2) Decidual seasonal forest (DSF): Occupying 6% of the AF, these forests prevail at higher altitudes where temperatures are lower. Like the SSF, they experience two distinct seasons—rainy summers and dry winters. Over 50% of their tree species shed leaves during the dry season. 3) Dense ombrophilous forest (DOF): This region occupies 17% of the AF, forming a verdant belt along the Brazilian coast. Characterized by the absence of a dry season, it experiences consistently high precipitation year-round and maintains temperatures averaging above 25°C. 4) Mixed ombrophilous forest (MOF): Commonly known as 'Araucaria forests', these forests are notable for consistent high precipitation and a diverse mix of evergreen and deciduous species. Covering 15% of the AF, they are distinguished by a high prevalence of gymnosperm species, notably Araucaria angustifolia. 5) Contact areas (CA): Covering 15% of the AF, these undifferentiated plant communities represent transitional ecotone areas, which occur at the boundaries or within the interiors of different ecoregions. In the AF, these ecotones can be found between forests and open physiognomies, such as those typical of the Cerrado, as well as between different forest types, like evergreen and seasonal forests. These areas are often characterized by a mix of vegetation types with varying degrees of canopy closure, reflecting the transitional nature of the environmental gradients they occupy. 49 Figure 1. Study area. Ecoregions within the Brazilian Atlantic Forest. Land cover maps To obtain maps of LULC types in the ecoregions of the AF, we used maps from the MapBiomas platform - Collection 8.0 (available at https://brasil.mapbiomas.org). The MapBiomas Project is a collaborative Brazilian initiative that maps LULCs across the entire territory of Brazil. It uses satellite images, specifically from the Landsat program with a 30-m resolution, along with machine learning tools to create an annual time series of maps from 1985 to the present (Alencar et al., 2022). Transitions between classes are defined based on detectable changes in LULC, such as shifts from anthropogenic to forest classes, identified through algorithms that analyze satellite imagery. The platform also captures reversals, such as deforestation events, and secondary vegetation, where secondary vegetation is converted back to anthropogenic use. Additionally, the platform provides information on the age of secondary forests (da Silva Junior et al., 2020). 50 We selected eight LULC maps, one for each five-year interval, spanning from 1986 to 2021 (Vancine et al., 2024). This interval helps minimize temporal autocorrelation, enhancing the independence of subsequent assessments of fire-LULC interactions. This same interval was previously used to assess vegetation dynamics in the AF (Vancine et al., 2024). Next, we simplified the 30 LULC classes from MapBiomas Collection 8.0 into seven categories (Table 1), plus a 'Noncombustible areas' type, which was omitted from further analysis. This reclassification was based on vegetation structural similarities and potential for comparable fire behavior, particularly in terms of fuel structure and composition (Nunes et al., 2005). Table 1. Characterization of land use and land cover types. Land cover type Definition Land cover type in MapBiomas Old-growth forests Areas with native forest cover since the beginning of the historical MapBiomas series. Forest Formation; Floodable Forest Secondary forests Areas that have transitioned from anthropogenic use to native forest vegetation. The transition period (vegetation age) can vary from 1 to 35 years. The captions for secondary forest areas provide information about the age of the vegetation. In this context, we grouped all age ranges into a single category corresponding to secondary forests. Openlands Areas characterized by ground and shrub vegetation. Savanna Formation; Herbaceous Sandbank Vegetation; Grassland; Pasture; Other Non-forest Formations Temporary crops Areas that are both sown and harvested within the same agricultural year, sometimes more than once. Soybean; Sugar cane; Rice; Cotton; Other Temporary Crops Perennial crops Areas with plant species that are cultivated and live for more than two years without needing to be replanted annually Coffee; Citrus; Palm Oil; Other Perennial Crops Planted forests Areas dedicated to silviculture, primarily consisting of Eucalyptus sp. and Pinus sp. Forest Plantation 51 Mosaic of uses Mix or combination of different land uses, creating a diverse and complex pattern across the landscape. It generally represents small areas that might include household agriculture and traditional cattle ranching (Almeida et al., 2016). Mosaic of uses Noncombustibl e areas Areas with low vegetation density (fuel accumulation) or high humidity and water levels. This class was excluded from the analysis. Mangrove; Hypersaline Tidal Flat; Wetland; Rocky Outcrop; Non vegetated area; Water Fire data To obtain maps of fires in the ecoregions of the AF, we used the MapBiomas platform – Fire Collection 2.0, which includes annual records of fire scars from 1985 to 2022 (MapBiomas, 2024). Similar to the LULC maps, this project also identifies burned areas through analyses of Landsat images with 30-m pixels, using the Normalized Burn Ratio (NBR) index (see more details in https://brasil.mapbiomas.org/metodo- mapbiomas-fogo/). To ensure greater accuracy in identifying the parts of the LULC that burned, we aligned our fire maps with our LULC data collection, which spans from 1986 to 2021 with a five-year interval, selecting fire maps from years that corresponded to one year after each LULC map. Consequently, our fire maps, also selected at five-year intervals, span from 1987 to 2022. The minimum fire extent was five hectares (see Pereira and Santos, 2003; Moreira et al., 2009; Silva et al., 2009; Carmo et al., 2011), with each fire assigned to an ecoregion of the AF. Data analysis To determine the relative fire-proneness of each LULC type, we employed a selection ratio-based approach. This method involved comparing, for each fire, the proportion of each LULC type within the burned area to its proportion in the total area 52 (burned and unburned) available for burning. In our study, a landscape unit is therefore defined as the spatial mosaic of different LULC types within the total area available for burning. To define the total area available for burning, several methods have been proposed in the literature. For example, Moreira et al. (2009) and Silva et al. (2009) defined circular buffers centered on each fire, with a radius corresponding to the size of the largest fire in the respective ecoregion, which was expected to approximate fire shapes in the absence of external influences. Alternatively, Oliveira et al. (2014) tailored buffers to the specific characteristics of each fire’s perimeter, using a shape that mirrored the fire but was approximately twice its size. Our approach aligns with Oliveira et al. (2014) in creating buffers that mirror the shape of each fire, but we opted for a buffer size of approximately four times the fire area based on preliminary analyses indicating that a larger buffer was necessary in our study region to capture variations in LULC while minimizing overlap between fires. This method was originally proposed for the study of resource selection by animals (Manly et al., 1993) and has been applied in fire research studies (e.g., Moreira et al., 2009; Silva et al., 2009; Carmo et al., 2011; Oliveira et al., 2014). The selection ratio (wi) for a given LULC type i is an index estimated as wi = oi/πi (Manly et al., 1993), where oi is the proportion of a burned patch belonging to LULC type i (estimated from the area consumed by fire) and πi is the proportion of available area belonging to LULC type i (estimated from both the burned patch and the surrounding buffer). Selection ratios range from 0, if the LULC type is available but not consumed by the fire (oi = 0), to a large value (in theory ∞), in the situation where the fire consumes the only tiny patch of a given LULC type that was available to burn (πi approaches 0) (Moreira et al., 2009). To avoid these extreme w values in very small areas, we only analyzed LULC patches larger than 1 ha. If a given LULC type is burned in proportion 53 to its availability, then w=1. If w>1, the LULC type is considered more fire-prone than expected by chance. Conversely, if w<1, the LULC type is considered less fire-prone than expected by chance. Since we were interested in the fire-proneness of LULC types per ecoregion, we omitted fires that, with their buffers, overlapped more than one region. A total of 40,869 burned patches were considered for analysis. Next, we averaged values for each LULC type across the fires within a given ecoregion and estimated 95% confidence intervals. We considered differences between selection ratios for different classes significant when their respective confidence intervals did not overlap. Additionally, if the confidence interval of w did not include 1, we considered the class significantly more or less prone to fire than expected by chance. Results Of the 40,869 fires detected and analyzed between 1987 and 2022, 44% occurred in the SSF ecoregion, 21% in CA, and the lowest percentages were in MOF (9%) and DSF (7%) (Table 2). The mean fire size was 23 ha, ranging from 15 ha in MOF to 27 ha in SSF. Mosaic of uses, secondary forests, and old-growth forests, in that order, were the LULC types with the highest available areas across regions. The region with the highest proportion of old-growth forests coincides with the area with the most fires, the SSF. CA had the highest proportion of secondary forests. 54 Table 2. Number of fires (≥5 ha), mean fire size (ha), and percentage of area occupied by LULC type, all with 95% confidence intervals, by ecoregion. LULC types Ecoregion No. of fires Mean fire size (ha) Old-growth forests Openlands Temporary crops Perennial crops Plantation forests Mosaic of uses Secondary forests Noncombustible areas Semideciduous seasonal forest 18287 20.88 ± 1.20 27.26 ± 0.04 11.04 ± 0.02 2.26 ± 0.02 1.37 ± 0.01 0.59 ± <0.01 33.34 ± 0.04 19.27 ± 0.04 4.88 ± 0.02 Decidual seasonal forest 2924 20.67 ± 1.50 18.80 ± 0.04 25.39 ± 0.06 0.02 ± <0.01 0.48 ± <0.01 0.86 ± 0.01 26.87 ± 0.06 26.25 ± 0.10 1.23 ± <0.01 Dense ombrophilous forest 7797 16.33 ± 0.70 22.76 ± 0.02 11.96 ± 0.02 2.39 ± 0.02 0.63 ± <0.01 1.46 ± 0.01 33.42 ± 0.06 21.20 ± 0.08 6.18 ± 0.02 Mixed ombrophilous forest 3472 13.91 ± 0.63 22.82 ± 0.06 9.97 ± 0.02 5.69 ± 0.02 0.62 ± <0.01 3.43 ± 0.03 29.03 ± 0.06 24.38 ± 0.10 4.07 ± 0.02 Contact areas 8389 16.79 ± 0.57 15.85 ± 0.02 13.76 ± 0.02 3.53 ± 0.02 1.51 ± 0.01 1.03 ± 0.01 21.14 ± 0.06 38.53 ± 0.12 4.65 ± 0.02 55 Fire-proneness across ecoregions Across the ecoregions of the Brazilian AF, secondary forests were the most fire- prone type of LULC, while old-growth forests were the least fire-prone (Fig. 2). Secondary forests burned 61% more than expected by chance (w = 1.61), whereas old- growth forests burned 57% less (w = 0.43). Besides secondary forests, only the mosaic of uses LULC type was significantly prone to fire, with a selection ratio 5% higher than expected. Like old-growth forests, the LULC types openlands, temporary crops, and planted forests were also significantly less prone to fire than expected, with selection ratios between 31% and 35% below expected. Perennial crops were the only type of LULC to burn according to availability across AF's ecoregions (confidence interval of w included 1). Figure 2. Mean selection ratios (w) with 95% confidence intervals for LULC types across ecoregions of the Brazilian Atlantic Forest. w = 1 indicates the LULC type burns according to availability; w < 1 means it burns significantly less, and w > 1 means it burns significantly more than expected by chance. 56 Fire-proneness by ecoregion As hypothesized, old-growth forests had low fire proneness across ecoregions (Fig. 3), burning between 33% and 66% less than expected by chance, depending on the region. Except for the DSF region, this LULC type had the lowest selection ratio in all other regions. In contrast, with selection ratios between 46% and 87% above expected, secondary forests tended to have the highest fire proneness across regions, except for the DOF. Openlands had low fire-proneness, burning 20% to 59% less than expected, except in MOF where they burned according to availability. As expected, other human-impacted LULC types showed variable fire proneness, depending on the ecoregion. Figure 3. Mean selection ratios (w) with 95% confidence intervals for LULC types in each ecoregion of the Brazilian Atlantic Forest. Refer to Fig. 2 for more explanations. 57 Fire-proneness over time The fire-proneness of old-growth forests remained consistently low over the 35- year study period (Fig. 4), showing no pattern of increase or decrease. Likewise, openlands and temporary crops generally had low fire-proneness, with no clear change over time, although temporary crops burned according to availability in 1992. Perennial crops, which generally burned in proportion to availability until 2012, became less prone to burning than expected from that year onwards. Forest plantations consistently had low fire-proneness, except in 1992 and 2002 when their fire-proneness was neutral. Mosaic of uses showed mixed fire-proneness over the years but shifted to a significantly lower than expected fire-proneness from 2012 onwards. It burned in proportion to its availability in 1992 and 2012, had high fire-proneness in 1987 and 1997–2007, but had low fire-proneness in 2017 and 2022. Secondary forests followed the hypothesized trend of decreasing fire-proneness, being significantly prone to fire until 2012, then burning in proportion to their availability from that year onwards. 58 Figure 4. Mean selection ratios (w) with 95% confidence intervals for LULC types across ecoregion of the Brazilian Atlantic Forest from 1987 to 2022. Refer to Fig. 2 for more explanations. Discussion Our analysis spanning over 35 years of fire-LULC interactions in the Brazilian AF ecoregions provides insight into the relative fire-proneness of different LULCs in a tropical biome. This is critical as fire becomes a growing issue both worldwide and in these biomes. Across all ecoregions, our results confirmed the low fire-proneness of old- growth tropical forests, in contrast with the high fire-proneness of secondary forests. Other LULCs showed more variable fire-proneness, with some differences between ecoregions. Furthermore, fire-proneness has changed over time in several LULCs, suggesting that these trends may be driven by natural ecological succession, climatic conditions, or human actions potentially influenced by changes in Brazilian environmental legislation. Notably, we found a decline in the fire-proneness of secondary forests over time, likely due to their lower susceptibility as the average age of these patches in the landscape increases, resulting in characteristics increasingly similar to old- growth forests. Fire-proneness in the Brazilian Atlantic Forest By confirming the low fire-proneness of old-growth tropical forests, our results are consistent with studies indicating a reduced risk and density of fires in undisturbed forests within the AF (de Assis Barros et al., 2022; Guedes et al., 2020; Singh & Huang, 2022). Despite the significant fragmentation and deforestation in the AF (Ribeiro et al., 2009; Vancine et al., 2024), we demonstrated that, in general, old-growth forest fragments still remain the LULC type least likely to burn compared to all others. The low fire-proneness of tropical old-growth forests can be attributed to their retention of transpired moisture 59 and the trees' substantial contribution to forest humidity, which limits fire ignition and spread (Cochrane, 2003; Cochrane et al., 1999). Interestingly, our data showed that the likelihood of old-growth forests burning less than expected across regions was similar to that of secondary forests burning more than expected, both at approximately 60%. The high fire-proneness of secondary forests was anticipated, as they exhibit high fuel load, low moisture content, and high radiation and temperature in the understory (Gutiérrez-Vélez et al., 2014; Hasselquist et al., 2010; Kyereh et al., 2007). The flammability of secondary forests—associated with their different species composition— may be influenced by the functional traits of their species (e.g., leaf thickness and surface area), which can change throughout the successional process (Parsons et al., 2015). Moreover, the spatial arrangement of these areas likely influences the fire spread, as they may be closer to anthropized areas and sources of ignition. This result is also consistent with previous works in the Amazon Forest that found higher fire activity in degraded forests (Balch et al., 2015; Brando et al., 2014, 2016; Cochrane et al., 1999). Research further suggests that, because secondary forests are often viewed as having low economic value, farmers may be less motivated to prevent them from burning when nearby fires pose a threat (Sorrensen, 2000). Other studies also show a link between deforestation fires to expand agricultural areas and the high fire-proneness of secondary forests (Pivello et al., 2021). Despite Brazilian legislation prohibiting fires, except in specific situations, farmers still burn secondary forests in pastures and agricultural areas to prevent forest encroachment. Furthermore, these areas are often re-cleared within the first few years following forest establishment (Piffer et al., 2022). Among the LULC types significantly less prone to fire, the results for openlands, where pastures are included, were noteworthy. Despite pastures having the largest burned areas in Brazil (Alencar et al., 2022; Moreira de Araújo et al., 2012) and being generally 60 associated with fire risk (Guedes et al., 2020; Gutiérrez-Vélez et al., 2014; Herrmann et al., 2023), our findings indicate lower-than-expected fire proneness. This shows the advantage of our selection ratio-based approach, as this method allows a distinction to be made between when a LULC type burns extensively due to its abundance in the landscape and when it burns disproportionately more than its availability. On the contrary, previous works have generally highlighted the large representativeness of pastures within the burned areas, overlooking the total available area that could have potentially burned in the landscape, which is usually also substantial. Indeed, although pastures occupy around 30% of the AF (Dos Santos et al., 2022; Souza et al., 2020), our results show that their fire proneness is likely lower than expected when availability per fire is taken into account. The high association of fire with pastures found in other studies may not be due to a positive selectivity compared to other land covers, but rather to the frequent illegal use of fire for cleaning and renewal pastures (Moreira de Araújo et al., 2012). It should be noted that although our results indicate lower fire proneness for pastures than expected by chance, pasture fires still pose a threat to the AF by contributing to the spread of fires to other fire-prone LULCs (Guedes et al., 2020). Management-related aspects of the other LULC types that were significantly less prone to fire, such as temporary crops and planted forests, or that burned according to availability, like perennial crops, should be stressed. Managed croplands usually exhibit low fuel load and high moisture content due to irrigation (Duguy et al., 2007). Planted forests, on the other hand, are often intensively managed by private companies to avoid fire hazards (Guedes et al., 2020; Mirra et al., 2017). Additionally, these LULCs are typically closer to urban areas, roads, and scattered populations. This proximity may result in more frequent fire occurrences (dos Santos et al., 2019; Barros et al., 2022), but it also facilitates faster fire detection and easier firefighting efforts, reducing burned 61 extent and fire proneness (Moreira et al., 2009). Lastly, it was to be expected that mosaic of uses would be prone to fire across ecoregions. These areas typically consist of small properties where slash-and-burn practices are employed to renew the land and eliminate waste. Subsistence farmers and indigenous populations often use these fires, confining them to small areas with localized impacts (Pivello et al., 2021). However, there is a significant risk that these fires could accidentally spread to nearby fire-prone LULCs. Under climate change scenarios, the risk of fire increases, especially when traditional agricultural fire practices result in accidents or loss of control (Pivello et al., 2021). It is crucial to develop strategies that balance traditional practices with modern fire management to mitigate these risks effectively. Fire-proneness by ecoregion Although it was hypothesized and therefore unsurprising that the old-growth forests of the AF exhibit consistently low fire-proneness with minimal variation between ecoregions, it is nonetheless remarkable that this low fire-proneness remains uniform across regions characterized by dry seasons, such as SSF and DSF, as well as in regions with evergreen vegetation, like ODF. These results reinforce the reduced fire-proneness of tropical forests (Cochrane, 2003; Cochrane et al., 1999), even under disparate regional conditions. Yet, previous studies have also shown that altitude, mean annual temperature, and drought severity can be positively associated with fire occurrence and lead to variations in the area burned in the AF (Abreu et al., 2022; de Assis Barros et al., 2022; de Santana et al., 2020; Herrmann et al., 2023; Singh & Huang, 2022). In the AF, regions with varying economic development can show different success rates in restoration programs (Piffer et al., 2022), affecting fire prevention and control. In the Brazilian Amazon, studies show that federal and state agencies, with distinct 62 institutional capacities, implement diverse fire management practices, contributing to regional differences in fire management (Fonseca-Morello et al., 2017). In Portugal, research also shows that regions with different climatic and environmental conditions present LULCs with different fire-proneness (Moreira et al., 2009). In our study, this is likely why the LULC openlands type, which generally had low fire-proneness in most AF ecoregions, exhibited higher fire-proneness, burning according to its availability, in the MOF region. This result is likely influenced by both the environmental and cultural context of the MOF ecoregion. This ecoregion is characterized by higher average altitudes and lower temperatures (IBGE, 1992), and is primarily concentrated in southern Brazil. Unlike the southeastern and northeastern regions, where intensive agriculture predominates, southern Brazil has a larger number of properties engaged in family farming. In these settings, the use of fire in pastures is a traditional practice (Carvalho & Andrade-Filho, 2019). Additionally, some southern states, such as Rio Grande do Sul, grant municipalities the authority to authorize and supervise the use of fire in pastures (Law No. 13,931, 2012). These regional differences are likely related to the smaller average fire sizes observed in this ecoregion compared to others. Despite the environmental, cultural, social, and economic differences between ecoregions, the secondary forest LULC type burned more than expected by chance in all AF regions, as predicted. Not even the consistently high rainfall in evergreen ecoregions such as ODF (Oliveira et al., 2000) reduces the fire-proneness of secondary forests, highlighting the vulnerability of this LULC to such disturbance. Still, our results show that secondary forests are almost twice as fire-prone in CA as in DSF, with selection ratios 87% and 46% higher than expected, respectively. This may be due to the fact that CAs are mainly transition zones between the AF biome and the Brazilian Cerrado. Unlike the AF, the Brazilian Ce