During 1988, the spatial variability of the natural gamma-emitting radionuclides (238U, 232Th, and 40K) and the artificial radionuclide 137Cs were studied in three near-level unfertilized fields in Saskatchewan, Canada. One field was used as a native (uneroded) control site; the other two sites have been cultivated, one since 1979 and the other since at least the 1940s. The two cultivated sites have been eroded by aeolian processes and tillage practices. Autocorrelation analysis indicated that most of the radionuclides were not serially correlated for the two depth increments sampled (0–15 cm and 15–30 cm). Autocorrelation is a necessary procedure when equally spaced transect sampling is employed, since sample independence cannot be assumed. Robust and nonrobust statistical summaries are presented for the selected gamma-emitting radionuclide concentrations. Robust statistical estimates of location and dispersion are favored when distributions are not normal or when the distribution is skewed. Median concentrations for the three fields indicated that 238U, 232Th, 40K, and 137Cs were similar to areas in the U.S. and to other areas in southern Saskatchewan. Median 238U activity concentrations were between 31.4 and 34.1 Bq kg−1; 232Th concentrations were between 29.6 and 31.2 Bq kg−1; 40K concentrations were between 471 and 502 Bq kg−1; and 137Cs concentrations were between 10.0 and 12.6 Bq kg−1. The variability of natural radionuclides in a given field for a specific depth increment was generally low, with coefficients of variation ≤ 10%. The variability of 137Cs concentrations was greater, ranging from 18% to 23%. Nonparametric tests indicated a significant decrease in 214Bi concentration (238U indicator corrected for 222Rn) with depth of the 1979 field, and an increase in 228Ac (232Th) with depth in the 1940s field. Significant differences in concentration between fields were indicated for all radionuclides except for 228Ac. The possible reasons for differences in depth and between fields are briefly discussed.
