Using a Non-contact Electromagnetic Induction Sensor to Measure Moisture Extraction By Root Systems of Sweetpotato Genotypes

We describe the use of an electromagnetic induction (EMI) sensor for proximal measurements of apparent electrical conductivity (ECa) to infer soil drying profiles in the root zone of sweetpotato breeding lines. Our hypothesis was that differential ECa readings during the storage root bulking stage (25 to 35 days after planting) was associated with variation in soil water depletion as a result of genotype-dependent variation in root development. We used a Geonics EM38 MK2 (Geonics Limited, Ontario, Canada) EMI sensor in horizontal dipole mode. In this study, we measured the ECa of 1,006 individual five-plant plots that were 1.5 m wide and 1 m long. The sensor was suspended about 5-10 cm above the canopy and individual plot measurements were obtained. The georeferenced data set was converted into a point shapefile using QGIS (v.2.10; Open Source Geospatial Foundation Project) and subsequently transformed into a raster layer using the inverse distance algorithm with default settings. The study area comprised 0.57 ha. After 120 days, the plots were harvested using a mechanical harvester and individual plots (genotypes) were selected based on predefined selection criteria. The location of each selected plot (SP) was georeferenced using a GPS-enabled Archer 2 ruggedized field computer (Juniper Systems, Logan, Utah, U.S.A.). GIS-based analysis indicated that 69% of the selections were obtained from plots that showed relatively low ECa (4.2-9.7), while the remainder were obtained from plots that showed relatively high ECa (12.27-14.7). These findings demonstrate the potential for the EMI-based approach to provide feedforward information about root development at the critical root bulking stage that can be incorporated into selection decisions at harvest. The raster model also showed a distinct moisture gradient in the field, indicating possible confounding effects associated with field variability. Such information can be used to reduce within-field variation, thereby improving selection efficiency.