The data of BSPs and exchangeable cation concentrations (H⁺, Al³⁺, Ca²⁺, Mg²⁺, K⁺ and Na⁺) are from Chinese Soil Database (http://vdb3.soil.csdb.cn/). After the comparison of data completeness and the elimination of abnormal data by the method of μ ± 3σ, 136 and 58 horizon samples from 59 and 28 soil species in red soil and yellow soil groups in South China were adopted, respectively (see Figure1), here, the spatial location of the selected soil species in Figurewere roughly determined according to the information of site description of the typical soil profiles in the available literatures [13]. Other main soil groups such as latosol and latosolic red soil groups were not considered due to the insufficient typical profiles (less than 8 soil profiles) after the elimination of abnormal data of BSPs and exchangeable cations.

According to the available historical records, the sampling and measurement of soil samples were as follows: 1) the location of each typical soil profile was mainly decided by the comprehensive consideration of the representation of topography, parent material, land use type, and etc.; 2) a standard soil profile

(1.0 m in width and 1.2 - 1.5 m in depth) was excavated; 3) the horizons were decided according to morphological characteristics of profile; 4) soil sample was collected uniformly within each horizon, and 2 kg of soil was kept left by using quartering and kept in the cloth bag. In the lab, soil samples were air dried, and the impurities of roots and gravels were removed. Then they were grinded to pass 1.7 mm, 0.25 mm and 0.149 mm sieves respectively.

Exchangeable K⁺ and Na⁺ were determined by flame photometer method, Ca²⁺ and Mg²⁺ determined by the method of EDTA volumetric method, H⁺ and Al³⁺ were determined by titration method after 1 mol/L KCl extraction, total amount of base exchangeable cations determined by the method of evaporation and neutralization titration after 1 mol/L NH₄OAc (pH 7.0) extraction, cation exchange capacity (CEC) determined by the method of 1 mol/L NH₄OAc exchange [8] [14] [15].

Microsoft Excel 2016 and IBM Statistics SPSS 20.0 were used for data processing, correlation modeling and mapping, difference significance of exchangeable cations and BSPs were tested by LSD method of one-way ANOVA, Pearson correlation (2-tailed) was used to describe the correlation between exchangeable cations and BSP [16] [17]. The contributions of exchangeable cations to BSP is calculated as follows: by using SPSS, the data of exchangeable cations and BSP were normalized, the multiple linear regression model between BSP and exchangeable cations was obtained, and then the percentage of the coefficient of i cation to total sum of the coefficients of all exchangeable cations in the model was calculated as the contribution of i cation to BSP [18]. Here, principle component analysis was not used to do so is because we want to quantitatively know the contribution of each cation to BSP.

Table 1 gives the statistical information of exchangeable cations and BSPs of red soil and yellow soil groups. It can be seen from Table 1 that except Na⁺, there were no significant differences in other exchangeable cations between red soil and yellow soil groups (p < 0.05). BSP of yellow soil group was significantly higher than that of red soil group (p < 0.05). cation concentrations of both red soil and yellow soil groups were in an order of Al³⁺ > Ca²⁺ > H⁺ > K⁺ > Mg²⁺ > Na⁺, in which Al³⁺ was significantly higher than all other exchangeable cations. For red soil group, Ca²⁺ and H⁺ were significantly higher than K⁺, Mg²⁺ and Na⁺, but no significant difference between Mg²⁺, K⁺ and Na⁺. For yellow soil group, Ca²⁺ was significantly higher than Mg²⁺, K⁺ and Na⁺, but no significant difference between H⁺, Mg²⁺, K⁺ and Na⁺. 4) All exchangeable cations and BSPs of red soil and yellow soil groups were in moderate variation (coefficient of variation < 100%) and in positive skew distribution (skewness > 0), while their probability density curves of all exchangeable cations and BSPs were near very flat (kurtosis = 0) or flat (0 < kurtosis < 0.67) [16] [17].

For the correlation between BSP with exchangeable cations, it can be seen from Table 2 that, 1) for red soil group, BSP had notable correlation (R = 0.5 - 0.8, the same below) with Ca²⁺, Mg²⁺ and Al³⁺ while low correlation (R = 0.3 - 0.5, the same below) with Na⁺ for red soil group; 2) for yellow soil group, BSP had notable correlation with Ca²⁺ and Al³⁺ while low correlation with Na⁺, Mg²⁺ and K⁺; 3) low correlation were found between H⁺ with Al³⁺, Ca²⁺ with Mg²⁺ and Na⁺ in both soil groups and between Mg²⁺ with K⁺ and Na⁺ in yellow soil group.

*. Different lowercase letters in the same columns show mean difference of exchangeable cations is significant at the 0.05 level, and different capital letters in BSP line show mean difference of BSPs is significant at the 0.05 level.

** and * Correlation are significant at the 0.01 and 0.05 levels (2-tailed).

Table 3 lists the optimal correlation models between BSPs with exchangeable cations and the multiple linear regression models between BSPs and exchangeable cations, it shows that most of the optimal models are in a quadratic pattern.

Table 4 shows the calculated importance of exchangeable cations to BSPs with the Random Forest Package, it can be seen from Table 4 that, 1) the importance of H⁺ and K⁺ of red soil were not significant to BSP, while only that of H⁺ not significant for yellow soil; 2) the orders of BSPs and the importance of the top three exchangeable cations to BSPs were different between red soil and yellow soil groups, which was Ca²⁺ (33.81%) > Al³⁺ (29.87%) > Mg²⁺ (22.28%) for red soil, while Al³⁺ (24.91%) > Ca²⁺ (21.55%) > Mg²⁺ (19.91%) for yellow soil; 3) although generally the importance of Ca²⁺, Al³⁺ and Mg²⁺ to soil BSPs are greater due to their high concentrations, no significant correlation was found between cation concentrations and their importance to BSPs if all exchangeable cations are considered.

Some studies already proved and explained that Al³⁺ was the main form in exchangeable acid exchangeable cations in soils of south China [19] [20] [21] [22]. Our data also showed Al³⁺ were the highest in all exchangeable cations both in red soil and yellow soil groups, which is common in soils of south China due to intensive desilicification but fersialitization caused by high temperature and precipitation [11] [12] [13], so it is normal for Al³⁺ as one of the important exchangeable cations controlling BSPs of red soil and yellow soil groups in south China.

* Importance is significant at 0.05 level; Norm. means normalized.

The lower concentrations of base exchangeable cations compared with Al³⁺ could be attributed to the heavy leaching losses of base exchangeable cations, which were higher than 80% [23] [24], due to the high temperature and precipitation in south China (annual temperature and precipitation as 16˚C - 25˚C and 800 - 2000 mm, dryness less than 1.0).

A significant difference was found in BSPs between red soil group and yellow soil groups, which could be attributed to the accumulation effects in the differences of the sums of acid exchangeable cations and base exchangeable cations between the two soil groups. It could be found that although there was significant difference in exchangeable cations, the sums of acid exchangeable cations and base exchangeable cations, the sum of acid exchangeable cations of red soil (4.77 ± 1.50 cmol(+)/kg) was higher than that of yellow soil group (4.41 ± 1.27 cmol(+)/kg), while the sum of base exchangeable cations of red soil group (0.69 ± 0.32 cmol(+)/kg) was lower than that of yellow soil group (0.74 ± 0.35 cmol(+)/kg). Generally speaking, red soil group was formed and evolved in the environment with higher temperature and precipitation (15˚C - 25˚C in annual mean temperature and 1200 - 2500 mm in annual precipitation) compared to yellow soil group (14˚C - 19˚C in annual mean temperature and 1000 - 2000 mm in annual precipitation), which resulted in higher acid exchangeable cations and lower base exchangeable cations of red soil group compared with yellow soil group [11] [12] [13].

Some studies found that BSP was dominated by Ca²⁺ in calcareous soil species on a small orchard scale due to the highest concentration of Ca²⁺ derived from the calcareous parent material [9] [10], but didn’t give information on the importance of exchangeable cations to BSPs. In contrast, our study not only found further that cation concentrations were in order of Al³⁺ > Ca²⁺ > H⁺ > K⁺ > Mg²⁺ > Na⁺ both in red soil and yellow soil groups (see Table 1), but also disclosed the importance of exchangeable cations to BSPs, for examples, the importance order of exchangeable cations to BSPs were Ca²⁺ (33.81%) > Al³⁺ (26.58%) > Mg²⁺ (19.82%) > Na⁺ (14.49%) > K⁺ (2.70%) > H⁺ (2.61%) for red soil group while Al³⁺ (24.91%) > Ca²⁺ (21.55%) > Mg²⁺ (19.91%) > K⁺ (14.24%) > Na⁺ (14.21%) > H⁺ (5.18%) for yellow soil group. However, our study shows that exchangeable cations controlling BSPs are different between red soil and yellow soil groups, we think such a complexity is normal because our study includes 59 and 28 soil species of red soil and yellow groups, respectively, which are widely distributed in south China with various soil forming environments and evolution degrees [11] [12] [13]. For examples, the spans of annual temperature and precipitation were 15˚C - 25˚C and 1200 - 2500 mm for red soil while 14˚C - 19˚C and 1000 - 2000 for yellow soil, respectively [11] [12], meanwhile, the parent materials include granite, sandstone, quaternary red clay, pelite, quartzite, tuff, basalt, shallow-sea sediment, andesite and limestone, etc. and the land use types include dryland, forest, orchard, shrub, grassland, etc. (see Table 5 and Table 6, only main parent materials and land use types presented in the tables),

*. Different lowercase letters in the same lines show mean difference is significant at the 0.05 level. Not included soil species with unclear information in parent materials or with the less numbers of profiles.

*. Different lowercase letters in the same lines show mean difference is significant at the 0.05 level. Soil species with unclear information in land use types were omitted in this study.

and it can be seen from Table 5 and Table 6 that Al³⁺, Ca²⁺ and H⁺ are the top three exchangeable cations in concentrations not only in different parent materials but also in different land use types, and there are certain differences in cation concentrations but no significant difference in BSPs between different parent materials or different land use types.

However, there are still some problems left to be solved in our study, for examples, why H⁺ is not significantly important to BSP with a higher concentration than K⁺, Mg²⁺ and Na⁺? Why K⁺ is significantly important to BSP of yellow soil group but not to red soil group?