Analysis of the spatial variability of maize root density

Abstract

A study was carried out in a maize field in order (i) to study the horizontal variability of the root length per unit volume (Lv), and (ii) to compare two methods of calculation of distances between roots: the first was based on the classical calculation of half the mean distance between neighbouring roots (HMDR), and the second was a direct graphical method carried out on root maps, on 5 superposed horizontal planes. Lv was measured at silking in adjacent, 10⁻³m³-sized parallelepipeds, and in 2-cm edged cubes taken in non-compacted zone. At a 10⁻³m³ scale, the distribution of Lv was bimodal in each layer, with one mode in non-compacted zones and the other, which was 20 times lower, below the wheel tracks. The study on a contimetre-sized scale in the non-compacted parts showed a very skewed distribution, with differences between cubes of more than one order of magnitude. The HMDR per layer were smaller than 2 cm to a depth of 60 cm, whereas the direct method showed that an appreciable proportion of points in the soil were at a distance more than 10 times the HMDR from the nearest root. It would therefore have been incorrect to consider, as it is generally assumed in water uptake models, that the maximum distance which water had to travel to the nearest root was the HMDR. Here, root mapping was therefore a better solution than measurement of Lv for characterizing root system as a water sink.