Climate and deforestation-induced changes in precipitation drive tropical forest-savanna transitions. However, precipitation alone provides a superficial understanding of the underlying mechanism behind these transitions. This is because our knowledge of how vegetation responds to changes in hydroclimate is fragmented. Under a rapidly changing climate, it is increasingly important to understand forest adaptation to predict future forest-savanna transition risks. However, there are two major bottlenecks to achieving this: (i) there is no universal metric that represents forest adaptation, and (ii) at continental scale, empirical evidence to ecosystem response under changing climate is still lacking. This thesis uses remote sensing-derived root zone storage capacity – a novel metric representing the vegetation's capacity to utilise subsoil moisture storage - and above-ground tree cover structure to provide empirical evidence to ecosystems’ response under changing hydroclimate and the influence of hydroclimatic adaptation on the resilience of tropical forests. The results reveal a non-linear relationship between ecosystem’s above-ground structure and subsoil moisture storage capacity. Furthermore, the ecosystem’s capacity to utilise subsoil moisture is much more dynamic and reflective of their transient conditions under changing precipitation than above-ground structure; thereby highlighting its application as an early warning signal. Ignoring this adaptive capacity can undermine forest resilience. The result from this thesis also emphasises the applicability of remote sensing in inferring and assessing ecosystem adaptation under rapid hydroclimatic change and can assist in strengthening management and conservation efforts across the continents.

Tropical rainforests in the Amazon and Congo River basins and their climate are mutually dependent. Evaporation from these forests help regulate the regional and global water cycle. Furthermore, these rainforests themselves depend on precipitation to sustain their structure and functions. However, the rapid increase in human activities (such as burning fossil fuels and deforestation) has significantly changed the rainforests’ climate. Due to the effect of human-induced perturbations on moisture feedbacks (i.e., precipitation and evaporation patterns), these rainforests risk tipping to a savanna or treeless state.

Understanding how these forests respond to climate change will aid in assessing their resilience to water-induced perturbations as well as in anticipating and preparing for potential tipping risks in the future. However, our understanding of how vegetation responds to climate change is fragmented, which limits our capacity to predict these risks. Previous studies have primarily relied on precipitation data to understand these forest-to-savanna transitions. However, ecosystem transition risks are also associated with water-stress, which depends on the vegetation’s capacity to adapt to drier conditions by storing water in its root zone. This thesis investigates the effect of hydroclimatic changes on root zone adaptation and its implications for forest resilience.

Paper I uses remote sensing data to analyse water-stress and drought coping strategies across the rainforest-savanna transects. Paper II uses the root zone storage capacity to quantify the resilience of forest ecosystems. Using the empirical understanding of root zone forest dynamics and hydroclimatic estimates from Earth System Models, Paper III projects future forest transitions and estimates tipping risks by the end of the 21^st century under four different shared socio-economic pathways. Paper IV uses atmospheric moisture tracking data to investigate the leverage landholders in South America have over precipitation and the resilience of forest ecosystems. 

Papers I and II reveal the non-linear relationship between the ecosystem’s above-ground structure and root zone storage capacity. These studies indicate that, under hydroclimatic changes, the ecosystem’s root zone storage capacity is much more dynamic than its above-ground forest structure and is more representative of the ecosystem’s transient state than precipitation. Ignoring this root zone adaptive capacity can underestimate forest resilience, primarily observed in the Congo rainforest. Paper III projects that the risk of forest-savanna transition will increase with climate change severity, most prominently observed in the Amazon rainforest. Paper IV finds that all landholders have equal leverage over the moisture precipitating locally and over farther-downwind land systems. According to this study, smallholders have a disproportionately larger influence over forest rainfall. However, large landholders have a larger influence on forest resilience as well as over the moisture precipitating on croplands and pastures. These results warrant the need for policies to factor in the impact of deforestation on downwind actors and promote effective ecosystem stewardship. The insights from this thesis highlight the importance of understanding and assessing ecosystem dynamics under a rapidly changing climate for strengthening management and conservation efforts across the globe.