Identification of earthworm burrow origins by near infrared spectroscopy: Combining results from field sites and laboratory microcosms

Abstract

A major obstacle to understanding feeding and burrowing behaviour of earthworms and their impact on soil structure in natural field conditions is that it remains impossible to identify the origins of burrows in the soil matrix. This gap in our knowledge makes it difficult to understand the roles of different earthworm species in creating the burrow network in soils. We tested the utility of near infrared spectroscopy (NIRS) as a new tool to overcome this obstacle. For the first time, we studied the ability of NIRS to detect specific chemical footprints left by different earthworms during the production of burrows, and thereby to identify the origins of burrows produced in the same soil matrix in the field, in terms of the earthworm species that produced them. The method was tested in three study sites (in sandstone, schist and marl geologies), differing in soil mineralogical composition and texture, with well-characterized and partially overlapping earthworm faunas. Burrows collected in the field were identified by comparing their NIR spectral signatures to the signatures of macroaggregates produced by the same earthworm species living in the same soil in laboratory conditions. We showed clearly that burrows of anecic and epi-endogeic earthworm species were characterized by specific NIR spectral signatures, resulting from quantitative and qualitative differences of OM in burrows among species. PLS-DA models conducted on NIR spectral data showed that NIRS technique allowed identification of a mean of 57.1% (±7.4) of burrows in soil monoliths at the sandstone site, 51.6% (±10.2) at the schist site and 46.5% (±8.1) at the marl site. Our study reveals some clear limitations in this method: low predictive abilities in surface soils, inability to discriminate among endogeic species and the requirement for a calibration procedure through the establishment of a site-dependent reference database.