Disentangling the long-term foliar 15N signal using a land surface model

Terrestrial vegetation growth is hypothesised to increase under elevated atmospheric CO₂, a process known as the CO₂ fertilisation effect. However, the magnitude of this effect and its long-term sustainability remains uncertain. One of the main limitations to the CO2  fertilisation effect is nutrient limitation to plant growth, in particular nitrogen (N) in temperate and boreal ecosystems. Recent studies have suggested that decreases in observed foliar N content (N%) and δ¹⁵N indicate widespread nitrogen limitation with increasing CO₂  concentrations. However, the factors driving these two variables, and especially the foliar δ¹⁵N values, are complex and can be caused by a number of processes. On one hand, if the observed trends reflect nutrient limitation, this limitation can be caused by either CO₂ or warming driven growth. On the other hand, it is possible that nutrient limitation does not occur to its full extent due to plant plastic responses to alleviate nutrient limitation, causing a decrease in N%, but changes in the anthropogenic N deposition 15N signal cause the observed δ¹⁵N trend. In reality, it is likely that all these factors contribute to the observed trends. To understand ecosystem dynamics it is important to disentangle the processes behind these signals which is very difficult based on observational datasets only.

We use a novel land surface model to explore the causes behind the observed trends in foliar N% and δ¹⁵N. The QUINCY (QUantifying Interactions between terrestrial Nutrient CYcles and the climate system) model  has the unique capacity to track ecologically relevant isotopic composition, including ¹⁵N in plant and soil pools. The model also includes a realistic representation of plant plastic acclimation processes, specifically a representation of nitrogen allocation to and inside the canopy in response to nitrogen availability, so implicitly to changes in CO₂  concentrations. We test the different hypotheses above behind the observed changes in N% and δ¹⁵N separately and quantify the contribution of each of the factors towards the observed trend. We then test the different hypotheses against existing observations of N% and δ¹⁵N from the ICP Forests database and other published datasets such as the global dataset of Craine et al. 2018.

Our study showcases the use of an isotope-enabled land surface model in conjunction with long-term observations to strengthen our understanding of the ecosystem processes behind the observed trends.