15N isoscapes in a subtropical savanna parkland: spatial-temporal perspectives

Abstract

Spatial patterns of soil δ15N reflect variation in rates of N-cycling processes across landscapes. However, the manner in which soil δ15N is affected by vegetation and topoedaphic properties under nonsteady state conditions is understood poorly. Here we propose and evaluate a conceptual model that explains how soil δ15N values will respond to changes in disturbance regimes (intensification of grazing and removal of fire) and the resultant invasion of a subtropical grassland by woody vegetation dominated by Prosopis glandulosa (honey mesquite), a N-fixing tree legume. Spatially-specific sampling along a catena (hill-slope) gradient where woody plants are known to have displaced grasses over the past 100 years revealed a positive relationship between soil δ15N and δ13C, and a negative relationship between NDVI and soil δ15N on upland portions of the landscape, indicating that plant cover is a critical determinant of δ15N spatial patterns. Because the dominant woody invader is a N-fixer, its invasion has increased N input and reduced soil δ15N. However, while honey mesquite also invaded and came to dominate lowland portions of the landscape, soil δ15N values in woodlands of intermittent drainages were significantly elevated relative to those in uplands. This is likely attributable to higher soil moisture, clay content, and total N in the lower portions of the catena gradient, which create conditions favoring more rapid Ntransformation rates, higher preferential 14N losses (e.g., gaseous), and thus enrichment of 15N. Thus, while spatial and temporal variation of soil δ15N has the potential to be an indicator of disturbance-induced changes in the net N balance, its sensitivity is compromised in topoedaphic settings with where rates of Ntransformation are high. Continued improvements in our understanding of controls over the spatial variability of soil δ15N at the landscape-scale will enhance our ability to use δ15N as a diagnostic tool for inferring N dynamics under both steady-state and disturbed conditions. © 2013 Bai et al.