Studying root distribution with geostatistics

Abstract

Geostatistics was applied to a set of RLD (Root Length Density) data of maize obtained from auger sampling. A reciprocal linear-logarithmic function was found to fit the spatial trend of RLD, and was subtracted from the original data for correct application of geostatistics. Subsequent use of variography enabled description of the relations of root density with space and provided information on the size of root clusters. Spatial relations also served as a basis for subsequent estimation by kriging of RLD in unsampled locations, thus yielding bi-dimensional RLD maps with higher accuracy than those obtained by traditional methods of interpolation. Comparing two maize hybrids of different yields, the maps showed substantial differences in the proportion of root length in peripheral positions of the cross-row vertical profile and in the incidence of zones with critical density of colonisation. The trend of variograms also indicated larger root clusters in the higher-yielding hybrid, together with more uniform colonisation of soil – a profitable feature in conditions of limited soil resources. The trial highlighted some drawbacks in the application of geostatistics to feasible schemes of soil-coring in wide-spaced crops.

Key words:

• Geostatistics
• ordinary kriging
• maize hybrids
• root clustering
• root maps

Notes

1. a, b and c, means of 4 replicates ± s.e.: − 0.642 ± 0.113, 0.317 ± 0.049 and 0.00632 ± 0.00168 respectively for DK585;  −0.697 ± 0.099, 0.333 ± 0.043 and 0.00998 ± 0.0211 for Santos.