Integrating soil physical and biological properties in contrasting tillage systems in organic and conventional farming

Though soil physical and soil biological properties are intrinsically linked in the soil environment they are often studied separately. This work adds value to analyses of soil biophysical quality of tillage systems under organic and conventional farming systems by correlating physical and biological data otherwise left unexplored. Multivariate redundancy analysis was used to relate data on soil water, soil structure, soil carbon, crop yield, and earthworm species abundances (Aporrectodea caliginosa, Aporrectodea rosea, Eiseniella tetraedra, Lumbricus rubellus). Structural equation modelling was then used to infer causal relations amongst the variables. Effects of tillage system (i.e., mouldboard ploughing (MP) and non inversion tillage (NIT)) on soil physical parameters and on the earthworm species Lumbricus rubellus were similar in organic and conventional farming. Despite sampling times in different seasons and different crops present at the time of sampling NIT correlated positively with L rubellus, soil organic matter content, plant-available water content, soil aggregate stability, soil water content, and penetration resistance. Field-saturated hydraulic conductivity was negatively correlated with NIT and was negatively, or not correlated at all, with earthworm species abundances, possibly due to the absence of Lumbricus terrestris in these fields. In the comparison of organic fields, earthworms were positively correlated with the soil's ability to hold water but loosening by ploughing appears to have benefited the conduction of water through soil more than earthworms. Tillage systems and farming systems were found to have both direct and indirect influences on soil parameters. Organic farming increased soil organic matter content, soil water content, and both endogeic and epigeic earthworm species abundances. Non-inversion tillage increased crop yield, soil organic matter content, and soil penetration resistance. This study demonstrates that multivariate techniques can integrate and add value to data otherwise analysed separately. (C) 2016 Elsevier Masson SAS. All rights reserved.