Résumé
A 2-years assessment of nitrogen (N) stocks, tree N safety nets (from a 15N labeling experiment), and biological N fixation was conducted in a Mediterranean agroforestry system in southern France. The study aimed to quantify N retention in agroforestry. The study area is characterized by a skeletic rhodic luvisol soil, a mean annual temperature of 15.5 degrees C, and an average annual precipitation of 556 mm. N and 15N were quantified across all system components, which are N-fixing black locust trees, crops, weed plants, understory vegetation strips (UVS), rhizospheric soil, and soil microbial biomass. A split-plot experimental design was used, with agroforestry (AF) and monoculture (MC) plots. 15N labeling was applied to track N fluxes and determine tree nutrient absorption over time. We hypothesized that tree root growth and nutrient leaching would enhance the safety-net effect, improving N retention. Results showed that crops in AF had lower N stocks (13-30% less than MC), but tree and UVS contributions compensated for the deficit, leading to a 62% increase in total plot-level N stock in 2022. The Relative Nitrogen Content was 0.97 in 2021 and 1.63 in 2022. While no 15N was detected in trees in the first year (70 days after labeling), 2% of applied 15N was recovered in trees 14 months post-labeling. % of N derived from atmospheric N2 (%Ndfa) ranged from 52 to 68%, with trees fixing 14-18 kg N ha-1. We confirmed that agroforestry enhances N retention, but further research is needed to quantify leaching and gaseous losses of nitrogen.