Electromagnetic induction (EMI), widely used in geophysical surveys over the last 80 years for obtaining subsurface soil information, has proved to be an efficient tool in terms of accuracy and ease of handling. However, the use of this indirect approach for mapping fields has been largely restricted to temperate agrosystems. The ability to map soil properties is essential for progressing towards the understanding of soil-vegetation relationships. Such relationships are of increasing importance in the tropics, where land-use changes tend to be more frequent and dramatic than in temperate regions. Hence, our investigation evaluated the use of EMI imaging for determining spatial variation in soil properties in a native tropical forest (NF) and a teak plantation (TP). We tested a combination of surface responses, using both direct sampling and a novel time-lapse approach, in an attempt to differentiate between variation caused by wetting patterns and variation caused by soil clay mineral patterns. Geostatistical analysis revealed a relatively constant soil electrical conductivity (ECa) spatial dependence of 31 m over both time (i.e. dry soils, wet soils) and sites (i.e. NF, TP). In addition, the results of a regression of EMI signal on the principal component scores of the main components for both sites revealed that the first component (accounting for 29% and 40% of the variance in NF and TP, respectively), explained a larger proportion of ECa spatial variation in TP (44%) than NF (23%). ECa was significantly correlated with several soil properties including hygroscopic water coefficient and content of clay or sand. Our results showed a more complex spatial pattern in NF than TP, with a short range of variation, suggesting the importance of the water repellency properties, pH and carbon content of the soil. This may provide a basis for future investigations of soil-tree relationships and soil property niche differentiation.