Applicability of Pedotransfer Models Between Yield Parameters and Some Chemical Soil Properties of Soybeanplant

Abstract

The aim of this study was to develop pedotransfer models on the basis of correlation between yield parameters of soybean (plant height, thousand seed weight, and seed yield) and some chemical characteristics of soils and to determine applicability of obtained models in estimation of plant yield grown in soils of Carsamba Plain. For this purpose, 40 soil and plant samples taken from the fields of the plain where soybean cultivation is carried out by the farmers were analyzed by random sampling method. The values of the yield indices of soybean plant, namely PH, TSW and SY varied between 172.33-351.22 cm, 167.54-450.75 g and 833.58-1584.37 kg da(-1) respectively. There was a significant correlation (0.476*) between PH value of soybean plant and TSW, as well as between PH value and copper (Cu) content (0.484**) of the soil. A significant positive correlation (0.384* ve 0.382*) was found seed yield value and potassium (K) and phosphorus (P) contents of the soil. Pedotransfer models between soybean plant height and soil properties of nitrogen (N), P, K, organic matter (OM), electrical conductivity (EC), cation exchange capacity (CEC), Cu, lime (CaCO3) high coefficient determination (R= 0.721) and was significant (p=0.078); Model between thousand seed weight and EC, OM, BD, calcium+ magnesium (Ca+Mg), N, P, K, CaCO3, sodium (Na), Cu, zinc (Zn), resulted in R=0.664 and was insignificant; and model between seed yield and EC, CaCO3, OM, N, P, K, Zn, Mn yielded R= 0.921 and it was highly significant (p= 0.000). Determination coefficient (R), root mean square error (RMSE), index of agreement (d), model efficiency (ME) were evaluated together to determine the validity of regression models. In determining the validity of the pedotransfer models obtained according to the experimental data, the data except the values used in the creation of the models were used. In general, statistical parameters were within validity limits. The results suggested that the developed regression models can be applied in the estimation of yield parameters in soybean grown in study area.