A solution for sampling position errors in maize and soybean root mass and length estimates

Root mass and length attributes are difficult to obtain in the field and currently there is uniformity among literature studies in estimating the effect of sampling position error. With the objectives of 1) quantifying the sampling position error in calculating weighted average root values per unit area and 2) developing an algorithm to minimize root position sampling error so that existing data in the literature can be used in future studies, we collected and analyzed root mass and length data across four sampling positions (0, 12, 24 and 36 cm distance from the plant row; row-to-row spacing 76 cm) from two maize and two soybean fields in central Iowa, USA. In-row sampling position (i.e., 0 cm from the plant row) over-estimated root mass and length by 66% and 46% for maize and soybean, while cores taken in the middle of plant rows (i.e., 36 cm from the plant row) under-estimated root mass and length by 34% and 23% for maize and soybean. As sampling distance from the plant row increased from 0 to 36 cm, maize root mass declined four times faster than soybean, while root length declined at almost the same rate between crops. Sampling 10 cm from the plant row provided the closest estimate to the weighted average value in both crops. We developed a new algorithm that predicts weighted average root attributes values with a R2 of 0.93 for mass and a R2 of 0.70 for length. The algorithm requires two user inputs (the measured root attribute value and the distance from the plant row). The new algorithm was tested across diverse environments, cultivars, and management practices and proven accurate for subsequent use (R2 = 0.70 and R2 = 0.87 for mass and length). This study provides guidance to strategically sample roots in future row crop research and an algorithm to eliminate sampling position bias in existing data.