Top-down and bottom-up drivers of soil delta15N spatial patterns in a tropical savanna

Although the stable nitrogen (N) isotope ratio (δ15N) can provide a powerful integrative tracer of highly localized nutrient cycling across heterogeneous savannas, the relative influence of top-down versus bottom-up drivers in shaping these spatial patterns remains largely unresolved. To address this gap, we leveraged long-term monitoring data alongside an extensive spatial sampling of 253 sites across a tropical savanna in Kruger National Park, South Africa. By applying a Generalized Additive Model (GAM) and variance partitioning, we evaluated the distinct contributions of top-down disturbances (fire and herbivory) and bottom-up drivers, including vegetation structure, soil properties, topography, and climatic variables. Our model explained 58.1 % of the total deviance and revealed high spatial heterogeneity in soil δ15N (ranging from 2.99 ‰ to 9.07 ‰). Variance partitioning demonstrated that bottom-up drivers overwhelmingly governed spatial variation, driven predominantly by the non-linear effects of highly abundant N-fixing trees, which dilute soil δ15N with depleted atmospheric N, and increasing soil C:N ratios, which slow mineralization and reduce fractionating N losses. Conversely, top-down drivers, while traditionally central to savanna ecology, played a comparatively minor role, primarily limited to the local enriching effects of extended fire-free intervals. Furthermore, the GAM model predicted isoscape highlights highly localized hotspots of 15N enrichment alongside broader spatial trends in soil δ15N that contrast sharply with total soil N. Ultimately, our study demonstrates that spatial variation in savanna soil δ15N is fundamentally dictated by localized bottom-up drivers and biogeochemical processes rather than top-down disturbances, highlighting the complex, multi-scale mechanisms shaping savanna N cycling processes.