Detecting Multiscale Carbon Controls in a Mediterranean Tree-Grass Ecosystem with Different Fertilization Treatments

Semi-arid ecosystems significantly impact the variability of the global terrestrial carbon sink and are influenced by a changing climate as well as anthropogenic nitrogen deposition. To understand how different nutrient availability impacts the drivers of net ecosystem exchange (NEE) in these ecosystems across temporal scales, we analyze a long-term dataset of biometeorological, soil and flux data from 2016 to 2022 from three Eddy Covariance (EC) measurement stations in Western Spain. The site (Majadas de Tiétar) is a Mediterranean tree-grass ecosystem home to a large-scale fertilization experiment. One station area received nitrogen treatment, another one both nitrogen and phosphorus treatments, and the third one serves as control. We apply Singular Spectrum Analysis to identify the variability of the different variables on multiple timescales (daily, multiday, seasonal). We then utilize metrics of mutual information to identify the main drivers of NEE across these timescales.

Preliminary findings show an increase in NEE dynamics in the fertilized areas both at ecosystem scale and at the grass-layer only. The dominant factors driving the dynamics of NEE vary depending on the timescale. On the daily scale, NEE is closely coupled to radiation, whereas on the seasonal scale water availability gains importance in predicting NEE. We find that the importance of vegetation greenness at the tree layer increases with longer timescales, and that also soil temperatures affect NEE at seasonal timescale. The relationships are further modified by different nutrient availability. While on the daily scale the differences between the three fertilization treatments are marginal, they intensify at the seasonal scale. Our comprehensive analysis facilitates a detailed understanding of the complex interactions between NEE and its controls under different environmental conditions and can help to improve the accuracy of terrestrial ecosystem models.