Tropical forests store more than a half of the world's forest carbon and produce over one third of the productivity of all terrestrial systems. They are also biodiversity hotspots, and host a large proportion of the world's terrestrial flora and fauna. However, growing evidence shows that the ability of tropical forests to perform important ecosystem services (i.e. carbon sequestration and biodiversity conservation) has been dramatically reduced by multiple pressures associated with human-induced forest disturbances (e.g. agriculture, logging, fire and fragmentation) and extreme climate events. Of these disturbances, fire represents of the greatest threats. Rainforests have not co-evolved with fire, and species have not adapted to withstand fire or the changes it imposes on the forests. Yet today, ignition sources are common in most human-modified regions, as many local farmers living within tropical forests traditionally use fire as a management technique to prepare their land for planting. This is compounded by selective logging and fragmentation, which increase the flammability of the remaining forests. Critically, fires are much more likely to escape their target area and enter the surrounding forests during severe drought events. This is exactly what happened during the current 2015-16 El Niño Southern Oscillation (ENSO) - considered one of the three strongest events ever recorded. The prolonged dry season allowed thousands of fires to get out of control in Amazonian and SE Asian tropical rainforests. Specifically in the Brazilian Amazon, the end of 2015 was marked by over 87,000 fire events, a 48% increase in relation to 2014 (a non-ENSO year). As a result, the widespread wildfires affected half of our 20 permanent plots near the Santarém region in the state of Pará, while fortunately preserving the other ten plots unburned. The Sustainable Amazon Network (SAN) has established these plots along a gradient of forest modification in 2010, and since 2014 a joint project between UK and Brazilian scientists (ECOFOR) has been carrying out research in this region. Consequently, the work we are proposing here benefits from unique and detailed pre-fire information on carbon dynamics and plant functional traits (from ECOFOR) as well as the distribution of three distinct taxa (birds, dung beetles and plants) and secondary seed dispersal processes (from SAN). Uniquely our network of permanent plots is established along an existing gradient of forest modification before the 2015 fires, allowing us to undertake the first rigorous evaluation of fire effects across different forest disturbance classes. This ability to examine fire impacts using detailed pre-fire data allows us to develop three major avenues of research across a human-modified gradient of forest disturbances: (1) the impacts of very severe wildfires on plant communities and carbon dynamics, assessing therefore which plant functional traits may predict species mortality, survival and recruitment; (2) an investigation into the fire impacts on forest fauna (i.e. birds and dung beetles) and associated seed dispersal processes; and (3) the development of a detailed understanding of scale and impacts of the current extreme ENSO-event, exploring the relationship between remote sensing information and ground-based measures. The better linkages between remote-sensing products and actual measures of fire severity will allow us to scale up the carbon emission and biodiversity loss estimates across the whole region. The results fo AFIRE are critically important, as tropical forests around the world may be threatened by drier, hotter and longer dry seasons with climate change. Our findings will help inform mitigation strategies to manage the impacts of future ENSO-mediated droughts and severe wildfires on tropical forests. We also expect AFIRE plots to form the basis of much longer-term research on the impacts of tropical wildfire

This proposal spans the three largest biomes in Brazil, the Atlantic and Amazon Forests, and Cerrado savanna. Together these cover >85% of Brazil's territory and include many of the most diverse ecosystems on Earth, but all have seen large losses in extent. While the value of their vegetation is increasingly recognized it is unclear to what extent these systems can regenerate or resist the increasing environmental stressors associated with climate change, particularly heating & drying. The motivation of BIO-RED is to understand how these changes affect the ability of intact & regenerating ecosystems to deliver societal benefits. This requires addressing these key questions: (i) How resilient are old-growth & regenerating ecosystems to the key stressors expected from future environmental changes? (ii) Is the destruction a reversible process on time-scales relevant to human society? Thus, will vegetation recover to a similar state as the original and provide similar services? (iii) Will the increasingly hot climate affect the recovery of forests and will modified forests be more vulnerable to future environmental change than intact forests? Answering these questions is only possible with a sound understanding how these systems function and what their sensitivities are. To respond to this need, BIO-RED will apply a multi-scale approach to evaluate the relationships between functions, biodiversity, resilience and regeneration potential in Brazil's three largest biomes in the face of deforestation and climate change threats. Our objectives are to: (i) Determine the biome-wide relationships between target ecosystem functions and biodiversity based on data from the RAINFOR and associated vegetation census networks; (ii) Obtain a detailed mechanistic understanding of the link between biogeochemical cycling, plant nutrient use and species composition and diversity in primary and regenerating systems at the local scale in 3 study landscapes; (iii) Examine tree species' ecophysiological sensitivities to key climate-linked stressors - drought, heat & fire - via real-time monitoring of vegetation functioning and comprehensive trait assessments; (iv) Develop and apply a UAV ("drone")-based imaging spectroscopy platform to map canopy chemistry and functional diversity at tree, plot & landscape scales, and explore the relationships between ecosystem properties & functional diversity; (v) Establish the extent to which biome transitions are already occurring, including forest invasion into cerrado, using both permanent plots and satellite-based monitoring. (vi) Determine the ability of recovering ecosystems and ecosystem management to protect biodiversity & provide key ecosystem services in Brazilian biomes; BIO-RED builds on existing observational networks all led by PIs of this proposal: RAINFOR, GEM, ForestPlots.net (>500 old-growth forest plots), ECOFOR & BIOTA, and others contributed by Brazilian project partners. Most activities will be focused on 3 focal-landscapes, in W Pará (Amazon forest), E Mato Grosso (cerrado), & E São Paulo (Atlantic forest), each with a complex mosaic of old-growth & regenerating systems that is already well sampled by our plot infrastructure and so ideal for intensive work to probe processes & to scale-up via hyperspectral imaging. BIO-RED will improve understanding of the extent to which Brazilian forest & savanna are resisting climate extremes, the extent to which destruction is reversible, & the vulnerabilities of intact & modified vegetation to climate extremes. It will identify the factors that control resilience & recovery of biodiversity & provision of key ecosystem services to people. These will be used to inform ecosystem management & policy options such as REDD+, the Brazilian Forest Code, & Brazilian ecosystem recovery plans. We therefore expect to lay a stronger scientific basis for future regeneration & protection of these systems, and so to improve benefits for human society.

The overall aim of this project is to determine and communicate the risk of significant change to the Amazon rainforest caused by anthropogenic disturbance and climate change. We will address a fundamental issue of our time, on the likelihood of Amazon rainforest dieback in the 21st century and identify regions that are most susceptible. We will combine this new knowledge with policies and scenarios developed by key stakeholders to co-design a Safe-Operating-Space for Amazonia. To address the iconic issue of Amazon dieback we will advance new ecological understanding of how forests grow, decline and recover following disturbance from climate extremes, forest fire and deforestation and their interaction in the context of 21st Century global warming. We will build novel datasets using a new forest plot network, drones and satellites to produce near-real-time maps of the risk to forests from climate, and track individual large-tree mortality across the basin. Together this information will be used in mathematical models to help estimate the risk of future forest dieback. We will join this work with models used to predict the effects of land use (forest conversion, degradation) on forest function, and the ecosystem services these forests provide to humanity. The outputs will enable us to deliver new information to policy makers regarding future options for land use, helping them to build optimal land use pathways that minimise the risks that may arise out of large-scale forest loss or dysfunction in Amazonia. The Amazon forest plays a vital role in the world's climate. In addition, by annually absorbing 5-10% of human-related CO2 emissions via vegetation growth, the region acts as a large brake on climate change. Climate extremes (eg drought), forest fires and deforestation reverse this process, causing net emissions to the atmosphere. If this were to happen on a large enough scale, via increased forest loss or increased rates of climate change - or their interaction - the resulting positive effect on global CO2 and climate change, would make the already-challenging Paris climate targets virtually impossible. In short, climate change, forest fires and deforestation have been identified as major intensifying and interacting threats to Amazonia. A substantive loss of Amazonian forest, also known as "Amazon dieback", would have huge negative consequences for human well-being, biodiversity, biogeochemical cycling, and regional and global climate. However, the level of global climate change combined with human disturbance that could trigger large-scale dieback is not known. Climate change is predicted to become more intense in the region alongside increases in human-driven deforestation and forest degradation (e.g fires, logging). Their impacts are poorly understood because of a lack of data, and because models cannot currently represent the key processes well enough. We have gathered leading UK and S American scientists in the fields of ecology, ecophysiology, Earth observation (using satellites) and the mathematical modelling of vegetation growth, land-use and climate as applied to Amazonia. We are uniquely positioned to make a step-change in understanding the combined effects of climate stress and human disturbance on Amazonia. Our measurements will build new knowledge about intact and disturbed forests, their stability and the physiology driving their stress responses. These knowledge advances will enable new modelling of forest-climate-land-use interactions which we will use to inform policymakers. We will engage with stakeholders from state to international levels to co-develop land-use scenarios that minimise risk in future climate and forest ecosystem services. Overall, we propose multiple large and integrated advances in empirical and modelling studies of the forests of Amazonia, and will build a science-policy dialogue that delivers significant impact locally, regionally and globally.