Spatial relationships among cereal yields and selected soil physical and chemical properties

Highlights: • Geostatistical analysis allowed delineating homogeneous low productive field area. • The low productive field area contains more sand and less silt and water in soil. • Spatial distribution of cereal yield and soil CEC, SOC and pH were related. • Kriging and remote sensing maps can help in upscaling management practices. Sandy soils occupy large area in Poland (about 50%) and in the world. This study aimed at determining spatial relationships of cereal yields and the selected soil physical and chemical properties in three study years (2001−2003) on low productive sandy Podzol soil (Podlasie, Poland). The yields and soil properties in plough and subsoil layers were determined at 72–150 points. The test crops were: wheat, wheat and barley mixture and oats. To explore the spatial relationship between cereal yields and each soil property spatial statistics was used. The best fitting models were adjusted to empirical semivariance and cross-semivariance, which were used to draw maps using kriging. Majority of the soil properties and crop yields exhibited low and medium variability (coefficient of variation 5–70%). The effective ranges of the spatial dependence (the distance at which data are autocorrelated) for yields and all soil properties were 24.3–58.5 m and 10.5–373 m, respectively. Nugget to sill ratios showed that crop yields and soil properties were strongly spatially dependent except bulk density. Majority of the pairs in cross-semivariograms exhibited strong spatial interdependence. The ranges of the spatial dependence varied in plough layer between 54.6 m for yield × pH up to 2433 m for yield × silt content. Corresponding ranges in subsoil were 24.8 m for crop yield × clay content in 2003 and 1404 m for yield × bulk density. Kriging maps allowed separating sub-field area with the lowest yield and soil cation exchange capacity, organic carbon content and pH. This area had lighter color on the aerial photograph due to high content of the sand and low content of soil organic carbon. The results will help farmers at identifying sub-field areas for applying localized management practices to improve these soil properties and further spatial studies in larger scale.