Accuracy of oil palm root length measuring methods

Abstract

Length is an important variable to describe root architecture and growth, but root length (RL) measurement is extremely laborious and time-consuming. Image analysis methods can provide reliable and fast RL estimates; such tools have not been applied for oil palm, a globally relevant commodity. We tested two image analysis softwares (Safira and ImageJ) and the modified line-intersect method to estimate oil palm RL. We considered the manual measurement of RL as reference to infer about the accuracy of the image analysis and line-intersect methods. We sampled roots in an 10-yr-old oil palm plantation in eastern Amazon. We used the estimates from the image analysis to calculate a correction factor that was further applied to improve the RL estimates. We applied the Paired sample t-test (p<0.05) to separately test the original and corrected estimates of each method in relation to the reference. The original RL estimates of the image analysis and line-intersect showed limited accuracy. The correction factors improved the estimates of RL for ImageJ, but not for Safira (except for roots within the 0.20-0.99 mm diameter range). We conclude that the ImageJ software is sufficiently accurate to estimate oil palm root length when specific root diameter correction factors are applied.