My research is interdisciplinary and focuses on understanding the long-term functioning and distribution of grass-dominated ecosystems. I use a combination of GIS, remote sensing data, and spatial statistical and theoretical modeling to study broad scale patterns of vegetation composition and structure as a function of climate gradients, disturbance regimes, anthropogenic impacts and resource availability. I also use paleoecology and historical ecology to study past responses of vegetation composition and structure to disturbances (especially fire), human impacts and climate change.

1- Calibrating sedimentary bio-proxies for quantitative reconstructions

Paleo-ecological reconstructions using bio-proxies conserved in natural archives give a unique opportunity to study vegetation dynamics on millennia scale and their responses to past climate and anthropogenic changes. However, to reconstruct vegetation structure and disturbances to units comparable with current and future measure of these variables, a calibration step is needed. Here are some examples of the calibration studies, improvement of bio-proxies analyses and taphonomy investigations:

2- Reconstructing past fire, herbivore populations and vegetation to understand the drivers of changes

The application of these calibrations to a paleosequence in Central African Republic (Lake Gbali), as well as the fossil pollen assemblage, showed that shorter period of climatic fluctuations were superimposed to this long-term trend, inducing shifts between savanna alternative stable states (Wonkam et al, submitted).
More long-term analyses are thus needed to better understand vegetation structure response to both human and climatic impacts. During the summer 2015, we (with Olivier Blarquez and Carla Staver) cored two lakes in the Congo. These sites will help us to disentangling the role of Bantu speaking peoples and abrupt climate changes three millennia ago in the formation and persistence of the periforest savannas. Indeed, their origins are still heavily debated!

3- Reconstructing past fire, herbivore populations and vegetation to inform conservation

Paleo-ecological reconstructions are critically needed to help conservation practices, in particular fire management in grass-dominated ecosystems. In 2015, we received funding from WWF to use sediment cores from the Mara Wetland, where the Mara flows into Lake Victoria, to understand how current ecosystem dynamics and rates of change compare to historical levels, and the effectiveness of conservation measures in the upper catchment. I am working with Amanda Subalusky, Chris Dutton, David Post, and Carla Staver (Yale), and Troy Hill (EPA), to analyze historical vegetation communities, fire regime and wildlife abundance in the basin. The results will enable us to understand how anthropogenic development in the basin has influenced sediment sources and mercury and nutrient levels in the river and to estimate the impact of conservation measures in the Mau Forest over the last two decades.

Similar work is ongoing in Saskatchewan, where native grasslands are one of the world’s most endangered ecosystems! Long-term data are needed to improve burning practices and grazing level, and to help understand the role of Aboriginal Peoples on the history of these grasslands.

Reconstructing past vegetation and biome location are also important to provide baselines for conservation. In particular, regarding forest and savanna location in the Tropics, this information is critical to prevent afforestation of savannas. We have a paper accepted dealing with forest extent in Africa since 1900, and some new estimates of deforestation rates.

4- Tropical ecosystems responses to global change

Savannas are present over large climatic and edaphic gradients, and experience frequent disturbances. This biome is thus one of the least deterministically predicted by climate, and questions remain on how it will respond to ongoing global change. I focused thus on identifying and quantifying the relative importance of the drivers responsible for savanna tree cover and savanna-forest transitions. By analyzing satellite and GIS data to model statistically tree cover, we demonstrated that fire-vegetation feedbacks through fire frequency and soil-vegetation feedbacks control savanna structure. By simulating future tree cover using the scenarios of climate and land-use developed by the IPCC for 2070, we showed that tree cover would experience drastic changes, leading to a lack of resilience for savannas of intermediate tree cover, especially due to land-use modifications.

5- Drivers of tropical biome distributions

To improve tree cover and biomes distribution predictions in the Tropics, we need a better understanding of ecosystems functioning at large spatial and long temporal scales. Theoretical modeling constitutes a good tool to test ecological hypotheses against current and past data. In collaboration with Carla Staver (Yale), we are developing new theoretical models of tropical savanna and forest. These models will be evaluated against modern, historical and paleo-data.