Unexplored transitions between savanna and forest in Africa

Abstract

Global environmental change is reshaping the composition, structure, and functioning of ecosystems worldwide. In tropical Africa, forests and savannas dominate the landscape and sustain the livelihoods of millions of people. However, if these ecosystems undergo significant shifts beyond critical thresholds, the consequences could be profound, affecting global carbon dynamics, biodiversity, and the provision of essential ecosystem services. This thesis aimed to explore the ecological assembly of forests and savannas across tropical Africa, providing insights into their current dynamics and potential future transformations.

In Chapter 2, I investigated landscape-scale patterns in Bicuar National Park, southern Angola, to identify the key drivers of community assembly within a mesic savanna ecosystem. Drawing on permanent plot data, functional trait analyses, and a phylogenetic framework, I showed that bottom-up environmental factors, particularly soil nutrient availability and catena position, play a dominant role in shaping metacommunity composition. I highlighted the influence of evolutionary history on the functional strategies of savanna species, demonstrating that traits that are evolutionarily labile emerge as the most critical in structuring these savanna communities.

In Chapter 3, I broadened the geographic scope to examine continent-wide patterns by using geo-referenced herbarium records to assess spatial phylogenetic alpha diversity across forests and savannas. I tested a long-standing hypothesis suggesting that transition zones—where forests and savannas converge—are floristically impoverished. My analyses revealed a decoupling of species richness and phylogenetic diversity in these transition zones, which harboured fewer species but greater phylogenetic alpha diversity than expected. I highlighted that these transition zones function as important centres of evolutionary innovation and diversification, either facilitating shifts between ecosystems or acting as evolutionary crossroads where lineages from contrasting environments coexist.

In Chapter 4, I analysed spatial patterns and drivers of phylogenetic beta diversity between forests and savannas across Africa. I found that savannas exhibited unexpectedly high levels of phylogenetic beta diversity compared to forests. Changes in phylogenetic composition in both ecosystems were structured along an East–West gradient, closely aligned with the Albertine Rift, and were primarily influenced by precipitation. I also identified the key plant lineages responsible for these broad-scale patterns in phylogenetic turnover. Finally, I demonstrated that, at the continental scale, environmental filtering exerted a stronger influence than dispersal limitation on the assembly of African plant lineages.

In Chapter 5, I addressed unresolved questions surrounding alternative ecosystem states theory by investigating whether transitional states between forests and savannas exist, and where they occur. Moving beyond a binary classification of forest versus savanna, I developed a continuous multivariate metric by integrating floristic, structural, and functional data from over 800 permanent plots across Africa. I then modelled this spectrum using remote sensing products and projected the results across tropical Africa. My findings revealed that intermediate states are widespread and have often been overlooked. I emphasised that transitional communities can manifest in multiple ways, and that a deeper understanding of forest–savanna dynamics requires greater attention to these transitional states.

This thesis expands our understanding of the processes that lead to the assembly of African tropical forests and savannas. It highlights the importance of incorporating multiple lines of evidence to dissect the ecological mechanisms that can lead to transitioning from one ecosystem to another. It also provides a nuanced view of whether forests and savannas exist as two discrete ecosystems, each a polar opposite of the other on an ecological seesaw, or whether intermediate states can exist. Future work will benefit from the continuous monitoring of permanent plots and the recording of unexplored biodiversity in remote areas to parameterise models based on new earth observation data.