Proximal and Remote Sensing Data Integration to Assess Spatial Soil Heterogeneity in Wild Blueberry Fields

Wild blueberries (Vaccinium angustifolium Ait.) are often cultivated uniformly despite significant within-field variations in topography and crop density. This study was conducted to relate apparent soil electrical conductivity (EC_a), topographic attributes, and multi-spectral satellite imagery to fruit yield and soil attributes and evaluate the potential of site-specific management (SSM) of nutrients. Elevation and EC_a at multiple depths were collected from two experimental fields (referred as Field_Und, Field_Flat) in Normandin, Quebec, Canada. Soil samples were collected at two depths (0–0.05 m and 0.05–0.15 m) and analyzed for a range of soil properties. Statistical analyses of fruit yield, soil, and sensor data were used to characterize within-field variability. Fruit yield showed large variability in both fields (CV_Und = 54.4%, CV_Flat = 56.5%), but no spatial dependence. However, several soil attributes showed considerable variability and moderate to strong spatial dependence. Elevation and the shallowest depths of both the Veris (0.3 m) and DUALEM (0.54 m) EC_a sensors showed moderate to strong spatial dependence and correlated significantly to most soil properties in both study sites, indicating the feasibility of SSM. In place of management zone delineation, a quadrant analysis of the shallowest EC_a depth vs. elevation provided four sensor combinations (scenarios) for theoretical field conditions. ANOVA and Tukey–Kramer’s post hoc test showed that the greatest differentiation of soil properties in both fields occurred between the combinations of high EC_a/low elevation versus low EC_a/high elevation. Vegetation indices (VIs) obtained from satellite data showed promise as a biomass indicator, and bare spots classified with satellite imagery in Field_Und revealed significantly distinct soil properties. Combining proximal and multispectral data predicted within-field variations of yield-determining soil properties and offered three theoretical scenarios (high EC_a/low elevation; low EC_a/high elevation; bare spots) on which to base SSM. Future studies should investigate crop response to fertilization between the identified scenarios.