pH is a basic but important property of soil, it can influence soil other physicochemical properties, microorganism activities and plant growth. For an acid soil (pH < 7.0), usually containing more potential exchangeable Al3+ and H+, pH measured by KCl extraction (here expressed as pHKCl) is conventionally used to indicate soil acidity together with pH measured by water extraction (here expressed as )  . Moreover, pHKCl is also used to identify Alice property (pHKCl ≤ 4.0) or Alice evidence (pHKCl ≤ 4.5) in soil taxonomy   .
It is well known that there is internal relationship between pHKCl and , but so far little information is available on the quantitative correlation between pHKCl and , possibly due to the easiness in measuring pHKCl as in the laboratory. However, it is still worthy predicating pHKCl directly from on the special conditions, for examples, when there is no information of pHKCl for soils in the historical literatures, or to save time and cost for a massive measurement in the laboratory. Meanwhile, some studies showed that soil pH is dominated by exchangeable Al3+, followed by the exchangeable H+ for acid soils in south China   , and exchangeable base, the contents of SOC and clays may also influence soil pH     , however, it is still unclear that the quantitative contribution of these influential factors to soil pH.
Thus, soil dada of Guangxi and Yunnan in South China obtained in 2014-2015 were used in this study in order to: 1) establish the optimal correlation model between pHKCl and , and 2) to quantitatively identity the contribution of exchangeable H+, exchangeable Al3+, exchangeable base, SOC and clay contents to soil pH.
2. Methods and Materials
2.1. Data of Soil Indexes
Soil Series Database of the National S & T Special Basic Project of Soil Series Survey and Compilation of Soil Series of China (Nos. 2008FY110600 & 2014FY110200) were used in this study. After the comparison of the data completeness of pHKCl, , exchangeable H+, exchangeable Al3+, exchangeable base, SOC and clay contents, and the elimination of abnormal data according to method of μ ± 3σ, 442 and 310 horizon samples from 126 and 98 soil profiles (0 - 120 cm in depth) surveyed from 2014 to 2015 in Guangxi and Yunnan (see Figure 1) were used to setup the optimal correlation model between pHKCl and , while 110 horizon samples from 30 soil profiles in Yunnan were adopted to quantitatively disclose the influence of exchangeable H+, exchangeable Al3+, exchangeable base, SOC and clay contents on soil pH.
2.2. Methods to Determine Soil Indexes
For soil indexes adopted in this study, and pHKCl were determined by potentiometer method after water extraction (soil: water is 1:2.5) and 1 mol∙L−1 KCl extraction, respectively. Exchangeable H+ and Al3+ by neutralization titration method after 1 mol∙L−1 KCl extraction, exchangeable base by drying neutralization titration method after 1 mol∙L−1 NH4OAc extraction, SOC by volume method after K2CrO4 digestion, and clay content by pipet method  .
Figure 1. Sites of soil profiles in Guangxi (left) and Yunnan (right) surveyed from 2014 to 2015.
2.3. Data Processing, Modeling and Mapping
Microsoft Excel 2016 and IBM Statistics SPSS 20.0 were used for data processing, modeling and mapping.
3.1. Statistical Information of Soil Indexes
Table 1 showed that soil pHKCl ranged from 2.41 to 6.51 for Guangxi and from 3.21 to 5.91 for Yunnan, with a mean of 4.27 for Guangxi and 4.20 for Yunnan, respectively, which prove further that the soils in South China are generally acid   .
It is well known that pHKCl is lower than because KCl could extract more Al3+ from soil particles than water  , however, Table 1 showed that the lower extents are different in different regions, for examples, pHKCl was 0.07 - 1.99 units with a mean of 0.99 units lower than for Guangxi, while 0.03 - 1.90 units with a mean of 0.89 units lower than for Yunnan.
Table 1 also showed that only of Guangxi presented as normal distribution (±0.1 < skewness < ±0.1), while of Yunnan, pHKCl of Guangxi and Yunnan as extremely-skewed normal distribution (skewness > ±0.3), The values of kurtoses of and pHKCl of both Guangxi and Yunnan ranged from −0.44 to 0.42, indicating the probability density curves of and pHKCl were very flat (kurtosis < 0.67)  .
3.2. Optimal Correlation Model between pHKCl and
IBM Statistics SPSS 20.0 is used to find the optimal correlation between pHKCl and . It can be seen from Figure 2 that: 1) The values of R2 showed again that there was significant positive correlation between pHKCl and for both Guangxi and Yunnan (p < 0.01). 2) The optimal correlation models were
Figure 2. Optimal correlation models between measured pHKCl and .
Table 1. Statistical information of measured soil pHKCl and .
different in Guangxi and Yunnan, quadratic model is found for Guangxi while exponential model for Yunnan, which may be attributed to the differences in climatic condition, soil types and land use types of the two regions. Values of MAE and RSME showed that the model accuracies were different in Guangxi and Yunnan, and the model accuracy is higher for Guangxi than Yunnan (Figure 3).
3.3. Threshold for pHKCl ≤ 4.0 or 4.5
It is our concern to know corresponding to pHKCl ≤ 4.0 or 4.5, which is the threshold of Alice property or Alice evidence in soil taxonomy. It can be seen from Table 2 that the estimated pHKCl would be most likely lower than 4.0 or 4.5 when the measured ≤ 5.0 or 5.5 for a soil in south China judged from the optimal correlation models in Figure 2.
Here, and pHKCl dada of 211horizons of 58 acid soil profiles surveyed from 2009 to 2010 in Guangdong were used to validate the proposed threshold (≤5.0 or ≤5.5), the results showed that pHKCl was within 3.08 - 4.21 (n = 173, pHKCl of 98.6% soil samples lower than 4.0) and from 3.08 - 5.00 (n = 202, pHKCl of all soil samples lower than 4.5) with a mean of 3.67 or 3.74 respectively when ≤ 5.5 or 5.0, indicating the proposed thresholds are reliable.
3.4. Contribution of Influential Factors to Soil pH
Table 3 showed the statistical information of values of pHKCl, clay, SOC, free Fe2O3, exchangeable H+, Al3+ and base of 110 horizon samples of 30 soil profiles
Figure 3. Comparison between simulated and measured pHKCl. MAE is mean absolute error, while RMSE is root mean square error.
Table 2. thresholds for pHKCl ≤ 4.5 or 4.0.
Table 3. Statistics of basic properties of soil samples of Yunnan (n = 110).
in Yunnan. Table 4 showed Pearson correlation coefficients between pHKCl with other soil indexes. It can be seen that there were extremely significant correlations between pHKCl with exchangeable H+ and Al3+, while significant correlation between pHKCl with exchangeable base, which indicate that exchangeable Al3+, H+ and base are the influential factors of soil pH, and their optimal correlation models with pHKCl were all in quadratic form (see Table 3 and Figure 4).
The linear regression model between pHKCl with exchangeable Al3+, H+ and base was obtained by using IBM Statistics SPSS 20.0 as pHKCl = 4.087 − 0.448H+ − 0.042Al3+ − 0.001 Base (R = 0.764, R2 = 0.583, F = 49.41, p < 0.01), which could be simplified as pHKCl = 4.087 − 0.448H+ − 0.042Al3+ since the exchangeable base could be neglected.
The linear regression model between pHKCl with the normalized exchangeable H+ and Al3+ was obtained as pHKCl = 4.064 − 0.211H+ − 0.522 Al3+ (R = 0.763, R2 = 0.581, F = 49.08, p < 0.01). The contribution of Al3+ and H+ to pHKCl were 71.1% and 28.7% after normalizing the coefficients of Al3+ and H+, which indicate further pHKCl is mainly dominated by exchangeable Al3+, followed by exchangeable H+  -  .
Figure 4. Optimal correlation models between pHKCl with exchangeable H+, Al3+ and base.
Table 4. Pearson correlation coefficients between pHKCl and other indexes (n = 110).
**Significant at 0.01 level, *Significant at 0.05 level.
Soil acidification is a natural process in tropical and subtropical regions due to higher rate of weathering and leaching under hot and humid climatic conditions    , it is accelerated greatly due to acidic deposition and excess application of acid fertilizer in agricultural system     . Our data showed that soil pHKCl ranged from 2.41 to 6.51 for Guangxi and from 3.21 to 5.91 for Yunnan, which proves further that the soils in South China are generally acid    .
Although it is well known that pHKCl certainly has inner relation with , so far there is no quantitative model between the two indexes. Our study not only proved further the quantitative relation between pHKCl and , but also established the correlation models and further found that the correlation models are different in different regions, it is useful to predict soil pHKCl directly from when there is no available information of pHKCl or to save time and cost for a massive measurement in the laboratory.
Our study found the optimal correlation models between soil pH and exchangeable Al3+ and H+ are in quadratic form, other studies showed the optimal correlation models between soil pH and exchangeable acid are different, which include power  or quadratic forms  . Different optimal models may be attributed to the differences in climatic conditions, soil types and land use types for the different studied regions.
Studies already proved that the main form of exchangeable acidity is found in the form of exchangeable Al3+ in acid soils and disclosed the mechanism of soil acidification      . Our study showed that the content of Al3+ of soil samples in Yunnan were 5.29 ± 2.93 cmol (+) kg−1, higher than H+, which was 0.25 ± 0.41 cmol (+) kg−1, meanwhile, exchangeable Al3+ has more significant correlation (R2 = 0.716) with soil pH compared with exchangeable H+ (R2 = 0.487), Thus, it is normal that exchangeable Al3+contributes more (71.1%) to soil pH compared to exchangeable H+ (28.7%).
By using the database of soil properties of Guangxi and Yunnan, this study discloses that pHKCl is meanly 1.0 unit lower than . There is significant positive correlation between pHKCl and , but the optimal correlation models are in quadratic or exponential forms for different regions. There are significant negative correlations between pHKCl with exchangeable H+ and Al3+, and exchangeable Al3+ and H+ contribute 71.1% and 28.7% to soil pH, respectively.
This study was supported by projects of the National Natural Science Foundation of China (Grant No. 41877008) and the National S & T Basic Special Foundation Project (No. 2014FY110200). We would like to express thanks to the contribution from all the participants in field soil survey and laboratory work.