Different shades of green : From quantity to quality for improving tropical forest restoration

Summary

Tropical forests are vital ecosystems that store carbon, regulate climate and water cycles, and harbor over half of all terrestrial biodiversity. They also support the livelihoods of over a billion people. Yet these forests continue to be cleared at alarming rates, mainly for agriculture and extractive industries. In response, forest restoration has become a global priority. However, not all forest recovery is equal. While the overall area of tree cover, defined here as forest quantity, is increasing in some regions, the quality of this tree cover, i.e. what kinds of forests are coming back, and permanence of these new forests remain uncertain. This PhD research addresses the urgent need to move beyond simply measuring forest quantity and toward understanding forest quality how long they persist, and what benefits they provide to people and nature. This thesis explores tropical forest transitions and restoration outcomes across different types of tree cover. Using a combination of satellite remote sensing, drone-based LiDAR, and ground-based forest measurements, I examined tree cover change and forest structure across multiple regions, with a specific focus on the Brazilian Atlantic Forest which is one of the world’s most fragmented and biodiverse tropical ecosystems. In Chapter 2, I disentangled patterns of forest loss and gain by separating mature forests, second-growth forests, and tree plantations across eight regions in Ghana, Mexico, Brazil, and Australia. While second-growth forests are increasing in area, they are often short-lived, i.e. cleared again after an average of just 10 years and thus limiting their contribution to biodiversity and ecosystem functions. Meanwhile, tree plantations often replace natural forests, raising concerns about how forest transitions are measured and interpreted. In Chapter 3, I explored how different forest restoration strategies, such as natural regeneration, mixed-species plantings, and monoculture plantations, can be distinguished using high-resolution UAV-LiDAR data. This method proved effective in capturing differences in forest structure, offering valuable insights for monitoring restoration outcomes and informing policy and management. However, this approach alone was not sufficient to fully separate similar forest types. In Chapter 4, I used structural causal modeling to assess how ecological and socioeconomic drivers influence both the quantity and quality of naturally regenerating forests in the Atlantic Forest of Brazil. I found that while factors like slope and proximity to forest remnants help explain where forests regenerate, other drivers resulting from human pressures, such as soil compaction and forest fragmentation, are critical for understanding the biodiversity and ecosystem functioning of these forests. Furthermore, I found that that key drivers for regeneration initiation are also those that shape its further development. Together, these studies show that forest recovery is not a straightforward process. Restoration outcomes vary widely depending on the restoration strategies, local context, and persistence of new forests. The findings underscore the importance of distinguishing between forest types and focusing on forest quality to better support biodiversity and nature contributions to people. To move toward meaningful and just forest transitions, we must prioritize not just where forests return, but how and for whom they are restored.