Individuals, populations and species are expected to move, adapt or die in response to climate change. However, there are big gaps in our understanding of how these processes play out, making it difficult to predict what will happen in the future. One gap lies in the contribution of individual animal behaviour to these responses. However, it has been difficult to include behaviour in predictions because it is unclear how it scales-up to influence larger, more visible ecological patterns. Animal behaviour has received relatively little attention as a potential mediator of community-scale responses because predictions tend to focus on larger-scale, more visible impacts, such as species range shifts and extinctions. However, behaviour could provide an important piece of the puzzle because it can be modified much more rapidly than physiological tolerance, providing an almost instantaneous buffer against negative effects of climate change. Rapid responses are particularly important to cope with severe short-term disturbances; for example, our research has shown that following the coral reef mass bleaching event in 2016, reef fish decreased their aggressive interactions three-fold to conserve energy in what became a suboptimal environment. Whilst behavioural change might not deliver a lasting solution, the buffer it provides could be critical in such cases to buy extra time for ecosystem recovery or for longer-term physiological adaptations to develop. Yet rapid behavioural responses could also create unintended side effects by disrupting the "rules of engagement" that underlie community organisation. Therefore, it is imperative that we close this knowledge gap to enable accurate understanding and future predictions of species responses to climate change. Our project addresses this challenge by combining existing empirical data collected before and 6-12 months after a natural disturbance experiment with targeted collection of new data 3-4 years after the disturbance, to create an unprecedented time series of behavioural observations and multiple metrics that describe the ecological community. In addition, we will create theoretical models to reveal if and how changes in aggressive behaviour can alter the interactions between individuals of different species, and how this can scale up to re-organise ecological communities. Finally, we will test how closely these theoretical predictions match the field data to establish for the first time whether re-organisation of ecological communities following disturbance is triggered by modified behaviour. Coral reefs offer an excellent model system to test these questions because they host an incredible diversity of fishes that fight aggressively for access to resources, which is thought to be an important process for structuring the wider reef fish community. In 2016, an extended El Niño event of unprecedented strength led to sustained increases in ocean temperature throughout the Indo-Pacific, causing mass coral bleaching and subsequent mortality across the tropics. Our existing data provides a baseline for (before bleaching), and quantifies the initial rapid changes in (6-12 months after bleaching), fish behaviour and the structure of ecological communities across multiple reefs. As a result, we now have a unique opportunity to use this rare, large-scale natural experiment to explore how behaviour mediates community shifts in a realistic setting by incorporating a longer-term perspective. By quantifying these impacts in multiple locations, we can be sure that any observed changes are driven by the bleaching event, rather than other environmental or geological differences between reefs. Our work will generate the first robust theoretical hypotheses and empirical evidence for how behaviour mediates the wider ecosystem. The results will enhance understanding and enable ecologists to incorporate behaviour into predictions of species responses to climate change.