Root estimation accuracy and sampling representativeness in relation to sample size in a subtropical evergreen broad-leaved forest: comparison between soil core and minirhizotron method

Optimal sampling strategy is vital in root studies because of the high degree of variability found in the belowground ecosystems. To inform the appropriate sample size, ninety-six cores and minirhizotrons in a subtropical evergreen broadleaved forest were used to examine diameter-related variation in the distribution of root length and volume, and assess the effects of sample size on estimation accuracy and sample representativeness. The variation coefficients of root traits estimated by the minirhizotron method were much higher than those estimated by the soil core method. Furthermore, minirhizotrons observed higher proportions of root length and volume in small diameter classes, reflecting thinner roots than those estimated by the soil core method. The Monte Carlo simulation analysis shows that, for the 0–5 mm diameter class, under a confidence level of 90% and estimation accuracy of 80%, the sample size for soil cores required to estimate the root length and volume are 5 and 12, respectively, and for minirhizotrons being 54 and over 96, respectively. The cumulative root diameter class curve analysis shows that the sample sizes required to encounter 80%, 90% and 95% of all diameter classes for soil cores are 3, 6 and 10, respectively, and for minirhizotrons being 38, 61 and 77, respectively. We conclude that due to the fractal distribution of root diameter, and the difference in sampling volume between different methods, larger sample sizes are required for minirhizotrons or larger-diameter roots than for soil cores or smaller-diameter roots to achieve better estimation accuracy and sampling representativeness.