*3.4. Relationship between EC<sup>e</sup> and 1:5 Soil to Water Extract Electrical Conductivity Methods*

In Table 4, the linear relationships between EC<sup>e</sup> and EC1:5, for all soil samples, determined by the three different methods are presented. Analysis of the results showed that each 1:5 soil to water extract electrical conductivity method is strongly related with EC<sup>e</sup> since R<sup>2</sup> values are high (0.953 < R <sup>2</sup> < 0.972) and RMSE are low (1.02 dS m−<sup>1</sup> < RMSE < 1.27 dS m−<sup>1</sup> ). It also appears that the linear equations showed small differences regardless of the EC1:5 methods for all soils examined. These data confirm the existence of a strong linear relationship when the range of EC<sup>e</sup> is relatively great (Table 1).

**Table 4.** Regression equations describing the relation between saturated paste extracts EC<sup>e</sup> and EC1:5 determined by three different methods with the coefficients of determination (R<sup>2</sup> ) and root mean square errors (RMSE) for all soil samples examined.


As shown in Table 4, the relationship EC<sup>e</sup> = fEC1:5 using the USDA method is similar to the corresponding one reported by Kargas et al. [7], (Table 1) for Greek soils since both the two equations have almost the same slope (6.61 and 6.53, respectively).

However, analysis of the results for soils with EC<sup>e</sup> < 3 dS m−<sup>1</sup> showed that a percentage of 70% of experimental EC<sup>e</sup> values were lower than those calculated by the equations presented in Table 4. For this reason, the data were separated into two ranges based on the threshold value EC<sup>e</sup> = 3 dS m−<sup>1</sup> to evaluate whether the relationship EC<sup>e</sup> = fEC1:5 is described by different equations as reported by other researchers [26,27].

The slopes of linear equation describing the relation between EC<sup>e</sup> and EC1:5 determined by three different methods, as well as the R<sup>2</sup> and RMSE for all soil examined for EC<sup>e</sup> < 3 dS m−<sup>1</sup> and EC<sup>e</sup> > 3 dS m−<sup>1</sup> , are presented in Table 5.

**Table 5.** Regression equations describing the relation between saturated paste extracts EC<sup>e</sup> and EC1:5 determined by three different methods with the coefficients of determination (R<sup>2</sup> ) and root mean square errors (RMSE) for all soil examined for EC<sup>e</sup> < 3 dS m−<sup>1</sup> and EC<sup>e</sup> > 3 dS m−<sup>1</sup> .


As shown in Table 5, for soils with EC<sup>e</sup> < 3 dS m−<sup>1</sup> , the slope of the linear equation between EC<sup>e</sup> and EC1:5 has different value depending on EC1:5 determination method used with the smallest and the highest values obtained by the NRCS and Loveday method. Also, the values of the slopes of linear relationships, for both EC<sup>e</sup> < 3 dS m−<sup>1</sup> and EC<sup>e</sup> > 3 dS m−<sup>1</sup> , differ significantly from each other since in the case of EC<sup>e</sup> < 3 dS m−<sup>1</sup> these values ranged from 4.68 to 5.46, while they ranged from 6.60 to 6.71 in the case of EC<sup>e</sup> > 3 dS m−<sup>1</sup> . In addition, for EC<sup>e</sup> < 3 dS m−<sup>1</sup> R <sup>2</sup> values are lower (0.537 < R <sup>2</sup> < 0.718)

than those ones (0.917 < R <sup>2</sup> < 0.942) observed for EC<sup>e</sup> > 3 dS m−<sup>1</sup> indicating a strong linear relation between EC<sup>e</sup> and each EC1:5 determination method. than those ones (0.917 < R2 < 0.942) observed for ECe > 3 dS m−<sup>1</sup> indicating a strong linear relation between ECe and each EC1:5 determination method.

in the case of ECe > 3 dS m−1. In addition, for ECe < 3 dS m−<sup>1</sup> R2 values are lower (0.537 < R2 < 0.718)

*Water* **2020**, *12*, x FOR PEER REVIEW 8 of 13

**Table 5.** Regression equations describing the relation between saturated paste extracts ECe and EC1:5 determined by three different methods with the coefficients of determination (R2) and root mean

**EC1:5**

**ECe < 3 dS m<sup>−</sup><sup>1</sup>**

ECe–Loveday method ECe = 5.46 EC1:5 0.647 0.123 **ECe > 3 dS m<sup>−</sup><sup>1</sup>**

ECe–Loveday method ECe = 6.71 EC1:5 0.942 1.580

ECe–NRCS ECe = 4.68 EC1:5 0.718 0.189 ECe–USDA ECe = 4.89 EC1:5 0.537 0.130

ECe–NRCS ECe = 6.60 EC1:5 0.934 1.710 ECe–USDA ECe = 6.60 EC1:5 0.917 1.800

As shown in Table 5, for soils with ECe < 3 dS m−1, the slope of the linear equation between ECe and EC1:5 has different value depending on EC1:5 determination method used with the smallest and the highest values obtained by the NRCS and Loveday method. Also, the values of the slopes of linear relationships, for both ECe < 3 dS m−<sup>1</sup> and ECe > 3 dS m−1, differ significantly from each other since in

**Methods ECe = f EC1:5 R2 RMSE (dS m<sup>−</sup>1)**

square errors (RMSE) for all soil examined for ECe < 3 dS m<sup>−</sup><sup>1</sup> and ECe > 3 dS m<sup>−</sup>1.

Comparison between the same methods for both EC<sup>e</sup> < 3 dS m−<sup>1</sup> and EC<sup>e</sup> > 3 dS m−<sup>1</sup> showed a difference between slopes ranging from 18.5% to 28.9%. Thus, in order to compare various equations describing the relationship between EC<sup>e</sup> and EC1:5, both the agitation method of EC1:5 determination and the range of EC<sup>e</sup> for which the equation has been proposed should be taken into account. Specifically, as shown in Table 5 and Figure 1, the relationship between EC<sup>e</sup> and EC1:5 determined by the NRCS method has a slope of 4.68 for EC<sup>e</sup> < 3 dS m−<sup>1</sup> and 6.60 for EC<sup>e</sup> > 3 dS m−<sup>1</sup> . The differences among the methods may be even greater if the soil contains gypsum or larger amounts of calcite than those observed in the soil samples examined. Comparison between the same methods for both ECe < 3 dS m−<sup>1</sup> and ECe > 3 dS m−<sup>1</sup> showed a difference between slopes ranging from 18.5% to 28.9%. Thus, in order to compare various equations describing the relationship between ECe and EC1:5, both the agitation method of EC1:5 determination and the range of ECe for which the equation has been proposed should be taken into account. Specifically, as shown in Table 5 and Figure 1, the relationship between ECe and EC1:5 determined by the NRCS method has a slope of 4.68 for ECe < 3 dS m−<sup>1</sup> and 6.60 for ECe > 3 dS m−1. The differences among the methods may be even greater if the soil contains gypsum or larger amounts of calcite than those observed in the soil samples examined.

**Figure 1.** Relationship between ECe and EC1:5 for NRCS extraction method. A: all soil samples, B: soil samples range ECe < 3 dS m<sup>−</sup>1, C: soil samples range ECe > 3 dS m<sup>−</sup>1. **Figure 1.** Relationship between EC<sup>e</sup> and EC1:5 for NRCS extraction method. A: all soil samples, B: soil samples range EC<sup>e</sup> < 3 dS m−<sup>1</sup> , C: soil samples range EC<sup>e</sup> > 3 dS m−<sup>1</sup> .

Similar results regarding to the effect of agitation method, the range of ECe and the gypsum content on equation describing the relationship between ECe and EC1:5 have been presented by other Similar results regarding to the effect of agitation method, the range of EC<sup>e</sup> and the gypsum content on equation describing the relationship between EC<sup>e</sup> and EC1:5 have been presented by other researchers [3,26,27].

researchers [3,26,27]. He et al. [27] proposed a quadratic equation as a more appropriate equation to describe the relationship between ECe and EC1:5when ECe values are lower than 4 dS m−1. The fitting of a quadratic equation to the data of this study for ECe < 3 dS m−<sup>1</sup> gave R2 values of 0.74, 0.57 and 0.66 and RMSE values 0.096 (NRCS), 0.124 (USDA) and 0.115 dS m−<sup>1</sup> (Loveday method), respectively. A comparison between these RMSE values and those of the linear relationships presented in Table 5, showed a significant improvement only in the case of the NRCS method. It should be noted that there is a He et al. [27] proposed a quadratic equation as a more appropriate equation to describe the relationship between EC<sup>e</sup> and EC1:5 when EC<sup>e</sup> values are lower than 4 dS m−<sup>1</sup> . The fitting of a quadratic equation to the data of this study for EC<sup>e</sup> < 3 dS m−<sup>1</sup> gave R<sup>2</sup> values of 0.74, 0.57 and 0.66 and RMSE values 0.096 (NRCS), 0.124 (USDA) and 0.115 dS m−<sup>1</sup> (Loveday method), respectively. A comparison between these RMSE values and those of the linear relationships presented in Table 5, showed a significant improvement only in the case of the NRCS method. It should be noted that there is a

**Table 6.** Regression equations describing the relation between saturated paste extracts ECe and EC1:1 determined by three different methods with the coefficients of determination (R2) and root mean

**EC1:1**

ECe–Loveday method ECe = 2.12 EC1:1 0.988 0.68

**Methods ECe = f EC1:1 R2 RMSE (dS m<sup>−</sup>1)**

ECe–NRCS ECe = 2.07 EC1:1 0.986 0.63 ECe–USDA ECe = 1.93 EC1:1 0.991 0.74

In Table 7, regression equations describing the relation between ECe and EC1:1 determined by three different methods are presented taking into consideration the threshold of ECe value 3 dS m−1. The results showed that the same trends were observed for R2 and RMSE values as in the case of the results of 1:5 ratio presented in Table 5. As regards to differences observed in the slope of linear

Table 6 shows the relationship between ECe and the three methods of determining EC1:1 for all soil samples examined. The results showed that the relationship is strongly linear in all methods examined (R2 > 0.986) and RMSE values are low (0.63 < RMSE < 0.74 dS m−1). The values of both R2 and RMSE indicate that this linear relationship reliably estimates the ECe. However, ECe = f EC1:1

significant difference in RMSE values presented in Table 4 compared to RMSE values whether we

use the linear equation or quadratic equation to ECe estimation for ECe < 3 dS m−1.

linear relationships have different f coefficient for each method.

square errors (RMSE) for all soil examined.

significant difference in RMSE values presented in Table 4 compared to RMSE values whether we use the linear equation or quadratic equation to EC<sup>e</sup> estimation for EC<sup>e</sup> < 3 dS m−<sup>1</sup> .
