Spatial distribution of soil chemical properties in an organic farm in Croatia.

Abstract

Soil pH, electrical conductivity (EC), organic matter (OM), available phosphorus (AP), and potassium (AK) are some of the most important indicators of soil fertility. These soil parameters are highly variable in space and time, especially in agricultural areas, with implications for crop production. The aim of this work was to study the spatial variability of pH, EC, OM, AP and AK using kriging and co-kriging methods in the Rasa River Valley (Croatia). As co-variates for each variable we considered the distance from the sea (DFS), distance from the river channels (DFC), pH, EC, OM, AP and AK. Only the variables with a significant correlation with the predictor were used as predictor variables. The results showed that soils of the study area had high pH, EC, OM and AK values and a low concentration of AP. The spatial variability was high for EC and low for pH levels. pH, EC, OM and AK had significant positive correlations. All these variables had significant negative correlations with AP. The exponential model was the best to model OM, AK and AP. Spherical and Gaussian models were the most accurate to model pH and EC. Spatial dependence was high for soil AK, EC and pH, and moderate for soil OM and AP. The incorporation of auxiliary variables increased the precision of the estimations. CoK_DFS was the best method to predict soil EC and AP, while Cok_EC, was better to estimate soil pH and Cok_pH and Cok_OM predicted soil OM and AK with the best accuracy. The maps produced with the best predictors showed that pH, EC, OM and AK had high levels in the northern and eastern parts of the study area. The opposite trend was identified in relation to the AP spatial pattern.