Xuancheng city is located in the south Anhui province (117˚58'E - 119˚40', 29˚57' - 31˚19'N) with a total area of 1.23 × 10⁴ km², the mean annual sunshine duration, temperature, precipitation, dryness and frost-free period is 1784 h, 15.6˚C, 1200 - 1500 mm, 0.68 - 0.90 and 240 d, respectively, the landform is complex and diverse, mountains, hills and valley basins are interlaced in the south with a mean altitude of 1200 - 1800 m, hills and downlands are distributed in the middle with a mean altitude of 15 - 100 m, while alluvial plains are dominant in the north with a mean altitude of 7 - 12 m. The tobacco-planting soils are mainly paddy soils mainly under tobacco-rice rotation. The annual fertilization is pure N 90 - 120 kg∙hm⁻² with N:P₂O₅:K₂O of 1:1.4 - 1.7:3.2 - 3.5 in the tobacco planting season while pure N 180 - 225 kg∙hm⁻² with N:P₂O₅:K₂O of 1:0.5:0.8 in the rice planting season. For acid farmland (topsoil pH < 5.5), about 1500 kg∙hm⁻² of 250 μm-diameter dolomite powders are applied before planting tobacco, which is proved to increase soil pH by one unit [8] .

Based on the database of tobacco farmers of Wannan Tobacco Ltd. Company in 2015, in total 124 typical tobacco-planting farmlands were determined according to their spatial distributions (see Figure 1 and Table 1), these farmlands belonged to different model tobacco farmers due to their excellent skills in flue-cured tobacco planting and great planting area (about 2 - 7 hm²) of tobacco.

Random multipoint method was used for the topsoil sampling (0 - 20 cm) in each typical farmland. Soil samples were air dried, ground and passed 1 mm and 0.25 mm sieves, then pH was determined by potentiometer, exc. Ca²⁺ and Mg²⁺ were by atomic absorption spectrophotometer after extracted by ammonium acetate [14] [15] [16] .

Table 2 is the grading standards of soil pH and exc. Ca²⁺ and Mg²⁺ for tobacco planting [4] [5] .

Data were analyzed with Microsoft Excel 2013 and IBM Statistics SPSS 20.0, the spatial distribution of topsoil samples was mapped on the platform of Esri Arc-GIS 10.3.

Table 3 shows the statistic information of topsoil pH and exc. Ca²⁺ and Mg²⁺ of typical tobacco planting farmlands in Xuancheng.

For soil pH, it ranged from the low grade to the high grade in the whole Xuancheng city, also ranged from the low grade to the high grade in Xuanzhou district, Langxi county and Jing county, meanwhile ranged from the low grade to the suitable grade in Jingde county, Guangde county and Ningguo city.

For soil exc. Ca²⁺, it ranged from the low grade to the high grade in the whole Xuancheng city, also ranged from the low grade to the high grade in Xuanzhou

district, Langxi county, Jingde county, Jing county and Guangde county, meanwhile from the low grade to the suitable grade in Ningguo city.

For soil exc. Mg²⁺, it ranged from the low grade to the high grade in the whole Xuancheng city, also ranged from the low grade to the high grade in Xuanzhou district, Langxi county, Jingde county, Jing county and Guangde county, meanwhile only concentrated in the low grade in Ningguo city.

Table 4 shows the sample number distribution in various grades of soil pH and exc. Ca²⁺ and Mg²⁺.

To the whole Xuancheng city, 37.10% (46 farmlands), 62.10% (77 farmlands) and 72.58% (90 farmlands) of the topsoil samples were in the low grades of pH, exc. Ca²⁺ and Mg²⁺, respectively, which indicate these tobacco planting farmlands still need continuously apply dolomite powders. 52.42% (65 farmlands), 25.00% (31 farmlands) and 20.97% (26 farmlands) of the topsoil samples were in the suitable grades of pH, exc. Ca²⁺ and Mg²⁺, respectively, which suggest these tobacco planting farmlands should postpone applying dolomite powders. 10.49% (13 farmlands), 12.90% (16 farmlands) and 6.45% (8 farmlands) of the topsoil samples were in the high grades of pH, exc. Ca²⁺ and Mg²⁺, respectively, which prove these tobacco planting farmlands should immediately stop applying dolomite powders.

The severest low pH phenomenon occurred in Langxi county, 72.73% of its farmlands were in the low grade, meanwhile, there was little difference in other regions in low pH proportions of their farmlands, ranged from 31.03% to 46.15% of their total farmlands. The severest high pH phenomenon happened in Jing county, 20.69% of its farmlands were in the high grade, followed by Xuanzhou district and Langxi county, 14.28% and 10.49% of their total farmlands were in the high grade, meanwhile no farmland in Jingde county, Guangde county and Ningguo city in the high grade.

The severest low exc. Ca²⁺ phenomenon occurred in Langxi county, 90.91% of its farmlands were in the low grade, followed by Ningguo city, 75.00% of its total farmlands were in the low grade. There was little difference in Jingde county, Jing county and Xuanczhou district in low exc. Ca²⁺ proportions of farmlands, ranged from 54.17% to 61.90% of their total farmlands. The severest high exc. Ca²⁺ phenomenon happened in Jing county, 20.69% of its farmlands were in the high grade, followed by Xuanzhou district, 16.66% of its total farmlands were in the high grade. There was little difference in other regions in low exc. Ca²⁺ proportions of farmlands, ranged from 4.76% to 9.09% of their total farmlands.

The severest low exc. Mg²⁺ phenomenon appeared in Ningguo city, all farmlands were in the low grade, followed by Guangde county, Xuanzhou district and Jing county, 84.62%, 76.19% and 72.41% of their total farmlands were in the low grade, meanwhile 54.55% and 57.14% of their total farmlands were in the low grade in Langxi county and Jingde county. No farmland was in the high grade in Ningguo city, while there was little difference in other regions in high exc. Mg²⁺ proportions of the farmlands, ranged from 3.45% to 9.52% of their total farmlands.

The optimal regression models between soil pH and exc. Ca²⁺ and Mg²⁺ were established by using IBM Statistics SPSS20.0 (see Figure 2), the models show that pH had an extremely significantly positive Napierian logarithm correlation with exc. Ca²⁺ and an extremely significantly positive power correlation with exc. Mg²⁺. The above equations prove that pH will increase fast initially but then increase slowly and tends to be stable with the increase of Ca²⁺ and Mg²⁺, such a phenomenon is conform to the results in other studies [17] [18] .

Exogenous Ca²⁺ and Mg²⁺ usually are input into the farmlands by fertilization and applying dolomite powders. According to the fertilization information [19] , about 2.2 kg∙hm⁻² and 7.4 kg∙hm⁻² of Ca²⁺ are annually input into the farmlands by fertilization during tobacco-planting season and rice-planting season, respectively, while no Mg²⁺ in these fertilizers. About 336 kg∙hm⁻² of Ca²⁺ and 195 kg hm⁻² of Mg²⁺ are input into the farmlands by applying dolomite powders 1500 kg∙hm⁻², thus, in total about 338.2 kg∙hm⁻² of Ca²⁺ and 202.4 kg∙hm⁻² of Mg²⁺ are annually input into the farmlands.

Ca²⁺ and Mg²⁺ usually are moved out of the farmlands by harvest of tobacco and rice. Tobacco leaves are harvested while tobacco stems are removed out from the farmlands to avoid disease spread. According to measured data of 62 tobacco samples collected from Xuancheng during 2007-2008, the average concentrations of Ca²⁺ and Mg²⁺ are 17.0 g∙kg⁻¹ and 2.8 g∙kg⁻¹, respectively with the total biomass of 4500 kg∙hm⁻² (tobacco leaves plus stems), thus, about 76.5 kg∙hm⁻² of Ca²⁺ and 12.6 kg∙hm⁻² of Mg²⁺ are annually removed out from the

farmlands. The average concentrations of Ca²⁺ and Mg²⁺ are about 0.3 g∙kg⁻¹ and 0.8 g∙kg⁻¹ in rice ears [20] [21] with the total rice-ear biomass of 7500 kg∙hm⁻², thus, about 0.2 kg∙hm⁻² of Ca²⁺ and 0.6 kg∙hm⁻² of Mg²⁺ are annually removed out from the farmlands, thus, in total, about 17.2 kg∙hm⁻² of Ca²⁺ and 13.2 kg∙hm⁻² of Mg²⁺ are annually carried out from the farmlands.

According to the above data, the net annual increase of Ca²⁺ and Mg²⁺ in the farmlands are 321.0 kg∙hm⁻² and 188.9 kg∙hm⁻², respectively, which prove the application of dolomite powders not only promote pH of farmlands, but also simultaneously increase exc. Ca²⁺ and Mg²⁺ of farmlands.

Over high pH and exc. Ca²⁺ or Mg²⁺ is unsuitable for tobacco planting, further data analysis roughly showed that the farmlands with 3 times or more application of dolomite powders usually are potential in the high grades of pH, exc. Ca²⁺ and Mg²⁺. So it is necessary to monitor the farmlands dynamically in order to decide whether continuous applying dolomite powders which should be dependent upon the specific conditions of individual farmlands.