The extent of soil erosion was estimated at six transects on terraces in the Mackenzie Basin using the Cs-137 technique to quantitatively estimate erosion since 1953. Mean Cs-137 areal activity over all sites was 3.51+/-9 Bq/m(2), compared with an input value of 422+/-63 Bq/m(2), and indicated a mean soil loss of 2.2 cm. Cs-137 areal activities exceeded the input value at few sites; most of the sites that did exceed the input value were tussock pedestals or well vegetated areas. There was a good correlation between decreasing vegetation cover and decreasing mean Cs-137 areal activity for Mackenzie soils, but not for Pukaki soils. However, there were strong contrasts in Cs-137 levels between bare sites (268+/-17 Bq/m(2)) and vegetated sites (418+/-15 Bq/m(2)) for Pukaki soils. Mean loss of Cs-137 from bare ground was 36%, equivalent to a soil loss of 3.9 cm. By contrast, vegetated sites showed no soil loss. The mean Cs-137 areal activity for all pedestals was slightly greater (457+/-25 Bq/m(2)) than the input value, while non-pedestalled sites had a mean of 326+/-9 Bq/m(2) (an average loss of 2.8 cm of soil). This suggested that vegetated areas, including pedestals, were stable or gaining soil, while bare deflated sites were losing soil. Topsoil depths tended to show similar trends to Cs-137 areal activity, with thinner topsoils occurring where vegetation cover was poorer. Pedestals had topsoils that were on average 4 cm deeper than non-pedestalled sites. The soil losses estimated using Cs-137, contrasts in topsoil depth between pedestals and non-pedestalled sites, and the trend for topsoil depth to decrease as vegetation cover decreased suggest erosional losses of a large proportion of the A horizon. The contrasts in topsoil depth are greater than the erosional losses measured using Cs-137, suggesting considerable erosion before 1953.
