Which part of the root system of corn (Zea mays L.) is visible at transparent surfaces?

Abstract

Roots are growing in soil and hence are not available for direct observation. This causes quantitative root studies to be difficult and time consuming. As an alternative to washing roots from soil often root boxes, minirhizotrones and rhizotrones are used (BoHM 1979, SMIT et al. 2000). However, one prerequisite for the quantification of root traits from these methods is that the visible part of the root system is a representative part of the total root system. The research reported here compares roots of corn (Zea mays L.) visible at transparent surfaces with roots washed from soil using the method described by INGRAM and LEERS (2000). In principle, roots are grown in a soil layer that is 6…7 mm thick and allows root observations and root length measurements after scanner images are taken from the transparent acrylic surface of the container. Two experimental approaches were used for comparison: (i) soil was compacted to soil bulk densities from 1.25 to 1.8 g/cm3, and (ii) different initial soil water contents were established in soil layers prior to planting. 15 days after planting root length was measured on scanner images with available software (Quacos Software, copyright the University of Georgia), and root length was determined for two diameter classes, i.e. 0 to 0.7 and 0.7 to 1.3 mm(corresponding to seminal and secondary roots) from washed out roots using WinRhizo Software (Regent Instruments, Quebec, Canada). The comparison of both measurements showed that: There was no identical relationship between visible and total root length of the diameter class 0…0.7 mm over the range found in the containers, neither for the soil water content nor the soil bulk density experiment. Hence, the length of secondary roots could not be predicted from observations at the transparent surface. For roots with diameters 0.7…1.3 mm both, for soil differing in water content and bulk density, visible and total root length showed a strong linear correlation. However, the slope of the regression line differed between the experi ments.