Comparison of log transformed and scaled cone indices

Abstract

Changes in soil properties with time and position make it difficult to analyze cone index data taken throughout a growing season. It was hypothesized that scaling would aid in interpretation. Cone indices were measured at 10 different dates over two growing seasons in conventional- and conservation-tillage plots in Florence, SC. They were measured at 0.05-m depth intervals to a depth of 0.55 m at spacings of 0.1 m across two 0.76-m wide rows of soybean or maize. Cone indices were scaled by subtracting each value by the mean and dividing by the range of cone indices for each date of measurement. This yielded an equal mean (zero) for each date with a unique distribution. Unscaled values were transformed by taking their logarithm to normalize the data. Log transformed data varied significantly with date of measurement but not between treatments. Scaled values did not vary with date of measurement but did vary between tillage treatments. Both unscaled and scaled cone indices varied significantly with water content. Scaling has the potential to improve the analysis of cone index data by reducing or eliminating some of the confounding treatment effects.