Les toitures végétalisées sont des écosystèmes urbains et construits en essor constant en France et dans le monde. Elles sont associées à plusieurs services écosystémiques tels que la limitation du ruissellement des eaux de pluie vers les canalisations, la réduction des effets d'îlots de chaleur urbains ou l'augmentation de la biodiversité en ville. L'amélioration de la quantité et de la qualité des services écosystémiques attendus nécessite de comprendre l'influence des interactions entre les composantes de la toiture végétalisée, à savoir la composition du sol, sa profondeur et la communauté végétale, sur les multiples fonctions écosystémiques associées. Pourtant, ces interactions n'ont jamais été étudiées dans le contexte des toitures végétalisées. A l'aide d'expérimentations en milieu contrôlé puis en conditions réelles sur une toiture parisienne, nous avons cherché à comprendre comment les interactions entre les composantes des toitures végétalisées influencent des fonctions majeures liées aux cycles biogéochimiques du carbone, de l'azote et de l'eau, ainsi qu'à la pollinisation. Nous avons mis en évidence une influence majeure des interactions entre type de sol, profondeur du sol, espèces de plantes et diversité végétale sur (1) le niveau de réalisation des fonctions écosystémiques ainsi que (2) les interactions entre ces fonctions. Nous avons montré que le choix des composantes d'une toiture pouvait conduire à des compromis entre services écosystémiques. Nous proposons des pistes de conception et de gestion pour obtenir des toitures végétalisées multifonctionnelles. / Green roofs are urban constructed ecosystems, associated with multiple ecosystem services, such as urban heat island and stormwater runoff mitigation or support for biodiversity. Enhancing the quality and quantity of expected ecosystem services requires to understand how interactions between substrate composition, substrate depth and plant community affect multiple ecosystem functions. However, such interactions have never been studied on green roofs. Using experimental approaches under controlled and real conditions on a Parisian rooftop, we focused on the influence of soil-plant interactions on key ecosystem functions related to carbon, nitrogen and water cycles as well as pollination. We highlighted that interactions between substrate type, substrate depth, plant species and plant diversity affect (1) the level of ecosystem functions and (2) interactions between functions. We found that the choice of green roof components could lead to trade-offs between ecosystem services. We propose general guidelines for the conception and management of multifunctional green roofs.

Agriculture significantly impacts the global carbon (C) and nitrogen (N) cycles through land use change, soil C loss, greenhouse gas emissions, and increased fixed-N availability. Agriculture occupies a third of the terrestrial biosphere, making understanding its impacts on the C and N cycles critical. I used a novel analytical tool (solid-state 13 C nuclear magnetic resonance spectroscopy) to characterize properties of the C and N cycles in agriculture, including biochemical responses to N fertilizer and agriculture gas fluxes. A central component of the C cycle is the rapid exchange of carbon dioxide (CO 2 ) and oxygen (O 2 ) between the terrestrial biosphere and the atmosphere. Gas flux O 2 /CO 2 ratios (oxidative ratio-OR) vary depending on ecosystem type, plant species, and nutrient status. It is necessary to constrain OR to assess the uptake of anthropogenic CO 2 by the terrestrial biosphere and ocean. I measured the OR of the top three crops in the United States (soybean, corn, and wheat) and found significant variability. I additionally tested the effect of N fertilizer application on corn ecosystem OR and on the difference between respiration and photosynthesis OR and observed no detectable changes. Conversely, soil organic matter OR is different from gas flux OR values, likely due to the influence of past land use and fractionation of OR during decomposition. I also analyzed how anthropogenic inputs to the N cycle (N fertilizer) and sustainable agriculture practices (cover crop) change plant biochemistry. This work has immediate implications for the biofuel industry. A central challenge to cropping for cellulosic ethanol feedstocks is the potential environmental damage from increased fertilizer use. I showed that yield increases in response to fertilization are not uniform across biochemical classes (carbohydrate, protein, lipid, lignin) or tissues (leaf and stem, grain, reproductive support). Heavy fertilizer application yields minimal grain benefits and almost no benefits in residue carbohydrates, while degrading the cellulosic ethanol feedstock quality and soil C sequestration capacity. Further cost analysis of these results showed that it is not cost-effective for farmers to apply high levels of N fertilizer, whether the crop is intended for food or fuel.