This study aims to improve estimates of carbon dioxide release and uptake in the North American Arctic and subarctic regions. Several modeling approaches were tested, showing that a better representation of sunlight and temperature effects on ecosystems leads to improved estimates. This work provides new perspectives to better assess whether these regions act as sources or sinks of greenhouse gases and how they may influence the climate system by amplifying or slowing global warming.

We show how climate change may reshape the world's major biomes, such as forests, grasslands, and tundra. Therefore, we combined dynamic vegetation models, machine learning and observation-based biome maps. We found that many biomes are likely to shift polewards as temperatures rise. Even under low-emission scenarios, large areas could change. These results help identify regions most susceptible to climate change and support global efforts to protect and manage natural ecosystems in the future.

In the tropical Northwest Pacific, a region of warm and nutrient‑poor ocean water, we investigated why oxygen consumption does not match the expected release of nutrients like nitrogen and phosphorus. We sampled from the surface to 2000 m and found that layering limits mixing, traps nutrients at depth, and leaves surface waters nutrient poor. Under these conditions, algae and bacteria recycle materials in ways that decouple carbon breakdown from nitrogen and phosphorus release.

Shell growth in bivalve aquaculture removes ocean alkalinity, leading to CO₂ emissions. Here, we explore a circular solution by crushing shell waste and distributing it into coastal sediments, where its dissolution releases alkalinity that can compensate these emissions. Laboratory experiments showed substantial shell dissolution and alkalinity production, highlighting the potential for aquaculture circularity under suitable sediment conditions.

Seasonal changes in forest behavior are often hard to detect in carbon data. We developed a wavelet‑based approach to uncover how three beech forests grow and absorb carbon through seasons. The forests showed distinct spring growth, summer slowdowns and long‑term trends. All forests were carbon sinks with managed forests strengthening over time while the old forest weakened. We show how seasonal dynamics explain long‑term changes and offer a new tool for tracking forest responses to climate.

We studied 24 years of carbon exchange in a beech forest to understand why its productivity rises, peaks and declines each at different times of the year. We mapped climate effects onto short biological windows and found that stored carbon supports early growth, a one‑week bud formation period can set next year’s peak and bursts of dry air drive mid‑summer drops. Hot summers even raised productivity when water was available. These insights help anticipate forest responses to heat and drought.

Shrubs need to be represented in climate models, because they shape the response of high-latitude regions to climate change. We added three shrub types to the ORCHIDEE (Organising Carbon and Hydrology In Dynamic Ecosystems) land surface model using Arctic observations. Including shrubs reduces simulated biomass and carbon uptake, improving model agreement with measurements. This is a simple, observation-based approach and an important step towards better representations of shrubs and their climate change response in models.

It is vital to monitor drought-related reductions in forest productivity over large areas. We tested the ability of 12 satellite variables to capture negative drought impacts on forest productivity across 18 European forest sites. Surface temperature worked best across all forest sites and droughts. After drought, forest productivity recovered faster than some commonly-used satellite variables, meaning use of these variables may underestimate forests’ resilience to drought.

Methane is the second most important greenhouse gas with high potential for short term reductions of human induced global warming. We model methane emissions from the most important and most uncertain natural source: wetlands. We investigate how a number of assumptions, including human impact on natural wetlands, influences the wetlands and their methane emissions. Of the tested influences we find the most important to be how humans are altering the soil surface.

We studied how blue carbon stocks and CO₂ and CH₄ fluxes vary at the landscape scale in tropical seagrass meadows. By testing six seagrass meadows across Singapore, we found that carbon stocks 0-25 cm and flux variability is not shaped by geomorphic settings, at least at the intermediate landscape scale. Our results support the inclusion of greenhouse gas fluxes in carbon accounting schemes.

We studied 800 kilometers of the lower Lena River in Siberia to understand how this large Arctic river releases climate-warming gases and carries natural elements from frozen landscapes to the ocean. Using boat-based measurements and water samples, we found that CO₂ release changes strongly over short distances, while CH₄ variations are very small. Most dissolved carbon stayed stable, and many metals were carried by tiny natural particles. Warming and thawing ground may shift these balances.

Permafrost thaw can lead to vegetation shifts and higher methane (CH₄) emissions in peatlands. However, plant belowground controls of these emissions are uncertain. We investigated how plant roots and rhizomes contribute to CH₄ emissions along a peatland permafrost thaw gradient. We found that low herbaceous plant root tissue density and high rhizome surface area were associated with higher CH₄ emissions. This indicated increased plant-mediated CH₄ transport with thaw.

Coastal wetland restoration often removes invasive plants, but its effects on deep soil carbon are uncertain. We compared plastic mulching and deep tillage for eighteen months in an invaded wetland. Mulching reduced soil salt, increased microbial biomass in deep soil, and coincided with lower carbon and iron-bound carbon through one metre of soil. These results show that restoration methods can affect hidden deep soil carbon, so monitoring should include the whole soil profile.

New satellite observations are used to improve our understanding of regional-scale vegetation. Innovative artificial intelligence methods are employed to combine the available information. The effectiveness of this approach is evaluated using independent observations. The results show that agricultural practices such as irrigation can be modelled more accurately.

N₂O is a greenhouse gas produced by microbes during nitrogen cycling. We studied two bacterial species and found that the isotope signatures of the N₂O they produced changed strongly depending on growth conditions and cell activity. Early during growth, the distribution of N isotopes within the N₂O molecule also changed unexpectedly. These results show that N₂O isotope signatures are more variable than previously assumed and depend strongly on physiology.

We surveyed bacterial communities in six water samples from four Colombian Andean highland lakes using 16S rRNA gene sequencing. Samples differed in taxonomic composition, diversity, and predicted functional profiles, and low-abundance genera of potential health relevance were detected. Given the limited sampling and lack of concurrent environmental measurements, these findings provide an exploratory baseline rather than evidence of specific environmental or anthropogenic effects.

We studied nitrogen use in Swiss agriculture for four decades using detailed data and modelling. We found that nitrogen use efficiency improved, while nitrogen losses decreased. Yet, modelling results suggested that cropland soils may lose nitrogen. The results show that policies implementation benefited productivity and reduced pollution. On the other hand, soil nitrogen balance should also be monitored, underscoring the need for integrated nitrogen management.

We studied how tree species and local soil conditions shape carbon storage in Mediterranean forest soils. By comparing mixed and single-species stands and sampling beneath trees and in nearby open gaps, we found that soils under canopies held much more carbon. Mixed stands stored extra carbon mainly in faster-cycling forms, while longer-lasting carbon was limited mostly by soil texture. This shows that forest structure can influence carbon storage in a warming, drier climate.