The Seasonal Variation Characteristics and Inuencing Factors of Soil Ecological Stoichiometry with Different Land Use Types in Karst Rocky Desertication Area in Southwest China

Aims Study the inuence of different land use types on the stoichiometric characteristics of soil C, N, P, and explore its response to environmental factors. Methods The research takes soils of 6 different land use types (shrub grassland, dry land, economic fruit forest, sparse forest, primary forest, and waste grassland) from the demonstration area for comprehensive control of karst rocky desertication in southwest China-Grand Canyon of Guanling Huajiang, Guizhou as the research object. Results The result shows that the C, N, and P contents of primary forest among the six different land use types are the highest, which are signicantly higher than that of shrub grassland, dry land, economic fruit forest, and waste grassland. The C and N contents of waste grassland are the lowest, which are signicantly lower than that of primary forest, sparse forest and shrub grassland, roughly represented as primary forest>sparse forest>shrub grassland>dry land>economic fruit forest>waste grassland. The C, N, and P contents show certain seasonal variations, which are higher in autumn than in spring. In addition, soil C, N, and P contents show a trend of decreasing with the increase of soil layer thickness, with obvious surface accumulation characteristics. Conclusions The soil C:N of spring and autumn in 6 different land use types are lower than the national and global averages, and the C:P and N:P are much higher than the average value of karst soils in the same area, indicating that the soil in the study area is relatively sucient in N but P is relatively lacking. Redundancy analysis shows that soil C, N and P in different land use types are affected. The main environmental factors for the change of stoichiometric characteristics are the human activity intensity, landform, and the slope. The research results will provide relevant theoretical support for the control of rocky desertication and vegetation restoration in southwest karst. − 1 , 0.52(cid:0)1.13g ·kg − 1 , respectively. Among them, in the 0(cid:0)5cm soil layer, the C, N, and P contents of the primary forest are the highest, which are signicantly higher than that of the other ve land use types. The C, N, and P contents of the waste grassland are the lowest, which are roughly expressed as primary forest > sparse forest > shrub grassland > dry land > economic fruit forest > waste grassland. The soil layer of 5(cid:0)10cm and 10(cid:0)20cm also show a similar pattern. The variation range of C, N and P contents in shrub grassland, dry land, economic fruit forest, sparse forest, primary forest and waste grassland in autumn are 19.71(cid:0)48.68g·kg − 1 , 1.74(cid:0)4.53 g·kg − 1 , 0.66(cid:0)1.32g·kg − 1 , respectively. In the soil layers of 0(cid:0)5cm, 5(cid:0)10cm, and 10(cid:0)20cm, the C, N and P contents of primary forest are the highest, which is signicantly higher than that of shrub grassland, dry land, economic fruit forest and waste grassland. The C and N contents in the waste grassland are the lowest, and the P content in the shrub grassland is the lowest, which is signicantly lower than that of the primary forest, sparse forest, and shrub grassland. In addition, the soil C, N, and P contents of 6 different land use types have a trend of decreasing with the increase of soil layers, with obvious surface accumulation characteristics, and the soil C, N, and P contents have signicant differences between 0 ~ 5 cm and 10 ~ 20cm soil layers. study of Bai Yixin al. At the same time, dryland soil has the highest C:N among the six land use types, which is signicantly higher than that of primary forest and sparse forest. Therefore, protection of primary forests and sparse forest should be strengthened to reduce the loss of nitrogen. The study also shows that the soil C:P in spring and autumn are 28.15(cid:0)46.4, 24.23(cid:0)46.17, and the soil N:P are 2.52(cid:0)4.15, 2.14(cid:0)4.06, which are much higher than the average level of C:P (19.77) and N:P (1.37) of karst soil in the same region, and the C:P and N:P of primary forest and shrub grassland are signicantly higher than those of waste grassland. This may be because the karst area is rich in rainfall and strong leaching. Compared with other ecosystems, the P content is lower, reecting a certain degree of P element limitation, which is the same as the study of Bao Gan et al. (2017).


Introduction
Karst landforms are various surface and underground landscapes and phenomena formed by various external forces such as soluble rocks' dissolution and precipitation, erosion and deposition ; Sheng et al. 2015). The total area of global karst landforms has reached 5.1×10 7 km 2 , occupying 10% of the total area of the earth ). The karst area in southwest China regarding Guizhou as the center has an area of more than 5.5×10 5 km 2 . It is not only the three largest karst concentrated distribution areas in the world with the most complete geomorphological types, but also one of the four fragile ecological areas in China ). The fragile ecological environment and the unreasonable socio-economic activities of human beings can easily cause vegetation destruction, soil erosion, decline or even loss of land productivity, the gradual exposure of rocks, and the evolution process or result of rocky deserti cation landscapes on the surface Chi et al. 2020). At present, there are more than 100 million people living in this area, with a large and concentrated population density. At the same time, this area is also one of the main areas of poverty population living in China. As a prominent contradiction between people and land is, the fragile natural environment and unreasonable human activities have exacerbated karst rocky deserti cation, which has become an important ecological problem that restricts the sustainable development of society and economy in the southwest karst (Wang et al. 2002;Zeng et al. 2018). For this reason, since the 1990s, China has successively initiated a number of major ecosystem projects of rocky deserti cation restoration and management, including major engineering measures such as closing mountains for afforestation and returning farmland to forests, and achieved certain results. However, the rocky deserti cation area after treatment still has a series of problems such as simple ecosystem structure, poor stability, and weak resistance (Xiong et al. 2012).
Soil is an important material component of the terrestrial ecosystem and the carrier of many ecological processes (Cai et al. 2020).
Elements such as C, N, P, and K in the soil are often essential elements for plant growth, which have an important impact on plant individual growth, population dynamics, community stability and succession (Wardle et al. 2004;Liu et al. 2005). Due to the dual interference of human activities and natural cycles, the nutrient elements in the soil can produce irregular dynamic changes, which affect the growth and development of plants and the material and energy cycle of terrestrial ecosystems (Gao et al. 2019). Due to the rugged and uneven surface of the karst area, the soil also has spatial discontinuity and complexity, so diversi ed land use types have been formed in the karst mountainous area (Wu et al. 2020). At present, although there have been a large number of reports on the ecological stoichiometry of C, N and P in karst soils (Zeng et  focus on the characteristics of soil nutrient content and its spatial variation in the process of rocky deserti cation succession ). However, there are relatively few studies on the seasonal dynamics of soil nutrient stoichiometric characteristics and its environmental in uencing factors, which severely restricts the in-depth study of soil stoichiometric characteristics in karst ecosystems.
Thus, this study takes the soil from karst rocky deserti cation area in southwest China as the object to study the seasonal variation characteristics of soil C, N, and P under 6 different land use types: shrub grassland, waste grassland, dry land, economic fruit forest, sparse forest, and primary forest. Then, discuss the impact of environmental factors (rocky deserti cation grade, landform, altitude, slope, aspect, rock exposure rate, human activity intensity) on it, in order to provide scienti c basis for the control of rocky deserti cation and vegetation restoration in southwestern karst rocky deserti cation areas.

Experimental site
The study area is located on both banks of the gorge of the Huajiang section of the Beipanjiang River at the junction of Guanling County and Zhenfeng County, Guizhou Province (25°38' 19″-25°41'32″N, 105°38'31″-106°40'51 "E). As a typical karst plateau canyon landform, it is widely distributed carbonate rocks, with a 45-1450m undulating terrain, relatively heightened 1000m. This area belongs to the subtropical dry-hot valley climate, with an average annual temperature of 18.4 ℃, and the average rainfall is 1100mm and mostly concentrated from May to September. The vegetation is subtropical evergreen and deciduous mixed broadleaf-conifer forest, growing mainly in yellow soil and yellow lime soil. Due to the fragile ecological environment and unreasonable human activities, the vegetation in this area is seriously damaged with a serious rocky deserti cation, and now it is mainly secondary vegetation. The wild vegetation includes Pinus massoniana, Ligustrum lucidum, Platycladus orientalis, etc., and the shrubs include Pyracantha fortuneana, Broussonetia papyrifera, etc., while arti cial forest includes Zanthoxylum bungeanum, Hylocereus undulatus, Lonicera Japonica, etc.

Experimental design
Carrying out eld surveys and combining Landsat's remote sensing image maps, selecting 6 representative land use types in the demonstration area: shrub grassland, dry land, economic fruit forest, sparse forest, primary forest, and waste grassland for sample setting (Table 1). Set three 20×20m squares in each land use type, and select 5 sampling points in each square according to the Sshaped layout method. In April and August 2020, each sampling point should be 0 5cm, 5 10cm, 10 20cm collected in 3 layers. During the sampling process, the litter on the surface of the soil layer was rst removed. Each sample would mix the soil of the same layer evenly to form a mixed soil sample number and bag it back to the laboratory. After natural drying, use a ball mill to grind the soil and pass it through a 100-mesh sieve for related chemical analysis. According to the analysis method of Soil Agrochemical Analysis (Lu 1999), the total C of the soil was determined by the potassium dichromate oxidation-external heating method, the total N was determined by the Kjeldahl method, and the total P was determined by the elimination molybdenum-antimony anticolorimetric method.  2015), the grade of human activity interference was also divided into no, weak, moderate, strong, and extremely strong, and the value was assigned in the order of increasing interference grade as 1-5. The landforms were divided into ve grades: valley, eroded gully, eroded steep slope, peak cluster platform, and eroded platform, with values of 1-5 in turn.

Data processing
Use Excel 2010 and SPSS 20.0 to perform statistical analysis on the data. Use one-way ANOVA to analyze the impact of different land use types on soil total C, total N, and total P indicators, and adopt LSD multiple comparison method for signi cance analysis, drawing by Origin 2018. Pearson correlation analysis was used to study the relationship between soil C, N, P and their ratios of different land use types and environmental factors, while the Canoco 5.0 software was used for further redundancy analysis (RDA) on the relationship between soil organic carbon, total nitrogen, total phosphorus and environmental factors. The data in the graph was the mean ± standard error.

Results And Analysis
Seasonal variation characteristics of soil C, N, and P contents in different land use types Figure 1 shows that the soil C, N, and P contents of 6 different land use types in the study area show certain seasonal variations, and the C, N, and P contents in autumn are higher than those in spring. The variation range of C, N and P content in shrub grassland, dry land, economic fruit forest, sparse forest land, primary forest and waste grassland in spring are 18.49 47.42g·kg − 1 , 1.61 4.06g·kg − 1 , 0.52 1.13g ·kg − 1 , respectively. Among them, in the 0 5cm soil layer, the C, N, and P contents of the primary forest are the highest, which are signi cantly higher than that of the other ve land use types. The C, N, and P contents of the waste grassland are the lowest, which are roughly expressed as primary forest > sparse forest > shrub grassland > dry land > economic fruit forest > waste grassland. The soil layer of 5 10cm and 10 20cm also show a similar pattern. The variation range of C, N and P contents in shrub grassland, dry land, economic fruit forest, sparse forest, primary forest and waste grassland in autumn are 19.71 48.68g·kg − 1 , 1.74 4.53 g·kg − 1 , 0.66 1.32g·kg − 1 , respectively. In the soil layers of 0 5cm, 5 10cm, and 10 20cm, the C, N and P contents of primary forest are the highest, which is signi cantly higher than that of shrub grassland, dry land, economic fruit forest and waste grassland. The C and N contents in the waste grassland are the lowest, and the P content in the shrub grassland is the lowest, which is signi cantly lower than that of the primary forest, sparse forest, and shrub grassland. In addition, the soil C, N, and P contents of 6 different land use types have a trend of decreasing with the increase of soil layers, with obvious surface accumulation characteristics, and the soil C, N, and P contents have signi cant differences between 0 ~ 5 cm and 10 ~ 20cm soil layers.
Seasonal variation characteristics of soil C, N, P stoichiometric ratios of different land use types It can be seen from Fig. 2  Among the 0 ~ 20 cm soil the layers, dry land has the highest C:N, which is signi cantly higher than that of primary forest and sparse forest. The C:P and N:P of primary forest and shrub grassland are the highest, which are signi cantly higher than that of waste grassland. In autumn, the stoichiometric ratios of soil C:N, C:P, and N:P of 6 different land use types vary from 9.28 ~ 14.22, 24.23 ~ 46.17, 2.14 ~ 4.06, respectively. In the 0 ~ 20cm soil layer, dry land has the highest C:N and C:P, which are signi cantly higher than that of economic fruit forest and waste grassland. Shrub grassland has the highest N:P, and waste grassland has the lowest N:P, which are signi cantly lower than the N:P for 5 other land use types. In addition, the soil C:N and N:P of 6 different land use types show a decreasing trend with the increase of soil layer. In spring, the soil C:P shows a decreasing trend with the increase of soil layer. In autumn, it shows a trend of rst increasing and then decreasing with the increase of soil layer. It shows that seasonal variations have a certain in uence on the soil stoichiometry characteristics of different soil layers.

Correlation analysis of soil ecological stoichiometric characteristics and environmental factors of different land use types
It can be seen from Table 2 that the soil C, N, P of different karst land use types and their stoichiometric ratios have obvious correlations with their environmental factors. In spring, the C, N, and P contents of various soil use types are positive correlated with altitude and aspect, and the C, N content and N:P are extremely signi cantly positive correlated with landform, and they are negative correlated with slope, rock exposure rate, human activity intensity, and rocky deserti cation grade. Among them, the human activity intensity is extremely signi cantly negative correlated with the C, N, P contents and N:P. In autumn, the landform is signi cantly positive correlated with soil N and P content, and negative correlated with C:N. Slope, rock exposure rate, human activity intensity, and rocky deserti cation grade are mostly still negative correlated with the C, N, and P contents of each soil use type. Among them, rock exposure rate and rocky deserti cation grade are signi cantly negative correlated with soil C and N contents, and human activity intensity is extremely signi cantly negative correlated with soil N and P contents.  Table 3, the importance of environmental factors to the soil C, N, P and their stoichiometric ratios in different land use types is ranked as follows: in spring, human activity intensity > landform > slope > rock exposure rate > rocky deserti cation grade > aspect > altitude, among which in uence of human activity intensity and landform on soil C, N, P and their stoichiometric ratios reach a signi cant level (P < 0.05), and the explanatory variable are 29.1% and 21.8%, respectively. In autumn, human activity intensity > landform > slope > rock exposure rate > rocky deserti cation grade > altitude > aspect, among which the in uence of human activity intensity, landform and slope on soil C, N, P and their stoichiometric ratios reach a signi cant level (P < 0.05), the explanatory variables are 19.9%, 18.7%, and 17.5%, respectively. From the biplots of RDA (Fig. 3), in spring and autumn, the arrow lines of human activity intensity, landform, and slope are the longest, showing that these three environmental factors can play a good role in explaining variation in soil C, N, P and their stoichiometric ratios, the same as the importance ranking results in Table 3. Among them, the angles between the human activity intensity, slope, rock exposure rate, rocky deserti cation grade and soil C, N, P, etc. are mostly obtuse angles, showing a negative correlation; the angles between landform, aspect, altitude and soil C, N, P are complementary angles, showing a positive correlation; and the arrow between aspect and altitude is the shortest, indicating that it has little in uence on the variations of soil C, N, P and their stoichiometric ratios.

Discussion
The in uence of different land use types on soil C, N, and P content As two important components of the terrestrial ecosystem, soil and plants carry out close material and nutrient cycles and are highly correlated. Soil not only provides nutrients for the growth and development of plants, but also affects the distribution and succession of plant communities, and plants will also improve the soil fertility status of the growing area through the return of root exudates and litter (Zhang et al. 2014;Yao et al. 2018). In this study, the C, N, and P contents of primary forests among the six different land use types are the highest, which are signi cantly higher than those of shrub grassland, dry land, economic fruit forests, and waste grasslands; the C and N contents of wild grasslands are the lowest, which are signi cantly lower than those of primary forests, sparse forest, shrub grassland; roughly expressed as primary forest > sparse forest > shrub grassland > dry land > economic fruit forest > waste grassland. This is . This may be due to the fact that the primary forest is protected by returning farmland to forests, human activities are less disturbed, and the surface is mostly dominated by tall trees. There is abundant litter returning to the soil, which is conducive to the accumulation of soil organic matter. However, the waste grassland is sparsely vegetation, single community and affected by the outside world, so the soil nutrient content is low. In addition, the study also shows that the soil C, N, and P contents of the six different land use types in the study area present certain seasonal variations, and the soil C, N, and P content in autumn is higher than that in spring. This is also the same as the research result of Zhu Shuying et al. (2005). Generally speaking, the temperature in summer and autumn is higher, the number of microorganisms is high with high activities, and the decomposition rate is fast, which is conducive to the accumulation of soil organic matter; the lower temperature in winter and spring inhibits the activities of soil microorganisms, which is not conducive to accumulate soil organic matter (Sun et al. 2020).
Affected by various factors such as climate, soil microorganisms, plant roots, and litter, the surface soil is prone to accumulation of nutrients, and the phenomenon of "surface accumulation" appears on the vertical section of the soil. In this study, the soil C, N, and P contents of the six different land use types have a trend of decreasing as the soil layer increases. They have obvious surface accumulation characteristics, and the soil C, N, P content between 0 5cm and 10 ~ 20cm soil layer have signi cant differences. This is also consistent with most research results ). At the same time, compared with other areas, the soil layer in karst areas is shallow, and the phenomenon of soil "surface accumulation" is more obvious. Therefore, protecting the existing primary forests and carrying out reasonable land use methods are of great signi cance to reduce soil erosion and water loss and increasing soil fertility in rocky deserti cation areas.
The in uence of different land use types on the stoichiometric ratios of soil C, N, and P The variations in the stoichiometric ratio of soil C, N, and P re ect the variation in the soil nutrient elements, an indicator to measure the composition and quality of soil nutrients (Fanin et al. 2013). In this study, the variation range of soil C:N in spring and autumn is 10.05 14.03 and 9.28 ~ 14.22, most of which are lower than the global average soil C:N (13.3), indicating that the organic matter in the study area decompose rapidly and the mineralization effect is obvious, and the soil nitrogen is relatively su cient, consistent with the study of Bai Yixin et al. (2020). At the same time, dryland soil has the highest C:N among the six land use types, which is signi cantly higher than that of primary forest and sparse forest. Therefore, protection of primary forests and sparse forest should be strengthened to reduce the loss of nitrogen. The study also shows that the soil C:P in spring and autumn are 28.15 46.4, 24.23 46.17, and the soil N:P are 2.52 4.15, 2.14 4.06, which are much higher than the average level of C:P (19.77) and N:P (1.37) of karst soil in the same region, and the C:P and N:P of primary forest and shrub grassland are signi cantly higher than those of waste grassland. This may be because the karst area is rich in rainfall and strong leaching. Compared with other ecosystems, the P content is lower, re ecting a certain degree of P element limitation, which is the same as the study of Bao Gan et al. (2017).
The in uence of environmental factors on the soil ecological stoichiometric characteristics in different land use types Soil is an important part of the terrestrial ecosystem, and its nutrient characteristics are jointly affected by many environmental factors such as vegetation type, climate, and land use type ). The 7 environmental factors in this study (human activity intensity, landform, slope, rock exposure rate, rocky deserti cation grade, slope aspect, altitude) are all related to the stoichiometric characteristics of soil C, N, P of different land use types. Among them, human activity intensity, landform, and slope are the main environmental factors that signi cantly affect the variations in the stoichiometric characteristics of soil nutrients in spring and autumn. Approximately, the soil C, N, and P contents are signi cantly negative correlated with human activity intensity, slope, while it is mostly positive correlated with the landform. Related studies have shown that strong human activities are one of the important driving forces for rocky deserti cation. Human activities such as farming, grazing, and cutting fuelwood have changed the nutrient cycle process of the karst ecosystem, which will cause a series of ecological problems such as soil erosion and aggravation of rocky deserti cation to varying degrees (Wu et al. 2020), the same as the research results of Wang Weiqi et al. (2010). At the same time, due to the rugged and uneven surface, large undulations, abundant and concentrated precipitation in karst mountainous areas, slope is an important driving factor for soil nutrient loss. In this study, slope is negatively correlated with soil stoichiometric characteristics, probably related to the nutrient distribution of different terrain parts. Compared with eroded steep slopes, the soil layers of the valley terrain have higher nutrient content and more obvious loss, which is the same as the study of Hu et al. (2020). In addition, the karst area in southwest China has a large area and extensive development of karst landforms, resulting in high spatial heterogeneity and discontinuous soil distribution in this area. The landforms have also become an important environmental factor that affects the variations in soil nutrient characteristics (Zhang et al. 2011; Zheng et al. 2020). The study also found that except the above three main environmental factors, rock exposure rate, rocky deserti cation grade, altitude, and aspect also have varying degrees of in uence on the soil stoichiometric characteristics in different land use types. It should be pointed out that climatic factors are also important factors affecting karst soil nutrient variations. In view of the fact that temperature, precipitation, humidity and other climatic factors are not involved in this experiment, relevant research will be carried out in the later period to improve the impact mechanism of environmental factors in karst areas on soil ecological stoichiometry of different land use types.

Conclusion
(1) Among the six different land use types, the C, N, and P contents of the primary forest are the highest, which are signi cantly higher than that of shrub grassland, dry land, economic fruit forest, and waste grassland. The C and N contents of the waste grassland are the lowest, which are signi cantly lower than that of the primary forest, sparse forest and shrub grassland, roughly expressed as primary forest > sparse forest > shrub grassland > dry land > economic fruit forest > waste grassland. The soil C, N, and P present certain seasonal variations, which are represented by the soil C, N and P in autumn is higher than that in spring. In addition, the soil C, N, and P contents of the six different land use types have an overall trend of decreasing with the increase of soil layers, which has obvious characteristics of surface accumulation.
(2) In the six different land use types, the C:N of spring and autumn soils are lower than the national and global averages, while C:P and N:P are much higher than the averages of karst soils in the same area, indicating that the soil in the study area is relatively su cient in N and P is relatively lacking.  Biplots of RDA for environmental factors with soil stoichiometric characteristics in different land use types