1. A horizon depth as ecological indicator for grassland: Mapping approach by using geophysical methods GRELLIER Séraphine (1), FLORSCH Nicolas (2), JANEAU Jean-Louis (1), LORENTZ Simon (3), PODWOJEWSKI Pascal (1) References: Wilcox B.P. (2002). Shrub control and streamflow on rangelands: A process based viewpoint. Journal of Range Management 55: 31 8-326. Hibbard K.A., Archer S., Schimel D.S., Valentine D.W. (2001). Biogeochemical changes accompanying woody plant encroachment in a subtropical savanna. Ecology 82: 1999–2011. 1. INTRODUCTION 2. METHODS Study area: Potshini catchment (near Bergville), representative of the KwaZulu- Natal Drakensberg foothills - 28 48' 37" S; 29 21' 19" E. In the Slingram EM38 apparatus, one coil serves as a transmitter and produces an alternative magnetic field in the ground. It induces an electric field ( electric field and magnetic induction). The later leads to a density current where σ is the conductivity. These currents produce a secondary magnetic field which is measured by using the receiving coil. Hence the secondary field reflect the conductivity. 1 IRD c/o School of Bioresources Engineering and Environmental Hydrology (BEEH), Rabie Saunders Building, UKZN, Box X01, Scottsville, 3209, South Africa. 2 UMMISCO/IRD, 32, avenue Henri Varagnat, 93143 Bondy Cedex, France; UPMC, Paris; Dept of Mathematics and Applied Mathematics, UCT, South Africa. 3 School of Bioresources Engineering and Environmental Hydrology (BEEH), Rabie Saunders Building, UKZN, Box X01, Scottsville, 3209, South Africa. Ecological and soil survey: Trees mapping and topography have been realized with DGPS (Leica). Grain size fractions (pipette method) were measured every 5cm until 65cm depth (always after reaching the B horizon). 3. RESULTS and DISCUSSION 4. CONCLUSION The high correlation between trees density and conductivity shows thresholds: between 26 and 34 mS/m trees density is higher. It is a first step to use this geophysical method as an ecological indicator in grassland. However we can not conclude about the direct relation between conductivity and trees density. The existence of low trees density between 26 and 34 mS/m indicate that other ecological parameters play a role in the trees distribution. The presence of a cattle track located on the west of the studied area (trempling and browsing effect) as well as topography and hydrology could also explain the low trees density in this intervals of conductivity values. This promising method using EM38 and measuring A horizon depth will have to be tested deeper with more ecological parameters in order to be validated as ecological indicator for grassland. Figure 1 Validation by using measurements on a gully face and close electrical sounding. Clay amount and electrical resistivity were both measured on an up-dated face. Bayesian method: The Bayesian inverse computation is based on: While data are Gaussian-like, one makes use of: Relative sensibility of the 3 modes as a function of depth (layered medium) Red: VDM Blue: HDM (normalized) Brown: VDM at 50 cm height (normalized ) Slingram method: The ground integrating probe response depends whether the dipoles are handled vertically (VDM) or horizontally (HDM): this provide two independent measurements. A third one is obtained by hanging the device 50cm above the ground. But using three three measures, one can retrieve the three parameters of a two layer shallow sub-surface: the two conductivities and the depth of the interface. Principle of the EM38 EM38 survey, vertical position on the ground (VDM). Figure 2 control nameEM38Sounding point 12526 point 23142 point 34943 point 423 Depth (cm) of transition of A and B horizons on 4 control points. All controls are well validated except point 2, which can be explained by a non-two layers structure at this point: heterogeneity of the grassland appears here. Very low conductivity ==> low tree density There is no high tree density where high and very low clay amount occurs The expanding grasslands in Southern Africa and all around contribute to agro-pastoral activities and to the evolution of the ecological quality of soils. In this context, grasslands are sometimes invaded by trees, which have eventually a strong impact on the ecosystemss (Hibbard et.al., 2001, Wilcox et al. 2002). The mapping of the A horizon, which is involved in the resistance to erosion, could be a relevant indicator of the determinisms and interactions contributing to asses soil quality and soil evolution in the landscape. Objectives here: to understand the potential involving role of the A and B horizons in the presence of invading trees (Acacia Sieberiana) by using geophysical methods (Slingram). The transition between the A and the B horizon (depth < 0.6 m) appears stiff in term of conductivity contrast (A being rather resistive while B is clayey and conductive). As a very robust inversion procedure, the Bayesian approach is efficient to map the A horizon thickness and both layer conductivities, and reveals large clay variations in the B horizon.