The characteristics of soil water cycle and water balance on steep grassland under natural and simulated rainfall conditions in the Loess Plateau of China

Summary

Large-scale vegetation restoration has been helpful to prevent serious soil erosion, but also has aggravated water scarcity and resulted in soil desiccation below a depth of 200 cm in the Loess Plateau of China. To understand the dynamic mechanism of soil desiccation formation is very important for sustainable development of agriculture in the Loess Plateau. Based on natural and simulated rainfall, the characteristics of soil water cycle and water balance in the 0–400 cm soil layer on a steep grassland hillslope in Changwu County of Shaanxi Loess Plateau were investigated from June to November in 2002, a drought year with annual rainfall of 460 mm. It was similarly considered to represent a rainy year with annual rainfall of 850 mm under simulated rainfall conditions. The results showed that the temporal variability of water contents would decrease in the upper 0–200 cm soil layer with the increase in rainfall. The depth of soil affected by rainfall infiltration was 0–200 cm in the drought year and 0–300 cm in the rainy year. During the period of water consumption under natural conditions, the deepest layer of soil influenced by evapotranspiration (ET) rapidly reached a depth of 200 cm on July 21, 2002, and soil water storage decreased by 48 mm from the whole 0–200 cm soil layer. However, during the same investigation period under simulated rainfall conditions, soil water storage in the 0–400 cm soil layer increased by only 71 mm, although the corresponding rainfall was about 640 mm. The extra-simulated rainfall of 458 mm from May 29 to August 10 did not result in the disappearance of soil desiccation in the 200–400 cm deep soil layer. Most infiltrated rainwater retained in the top 0–200 cm soil layer, and it was subsequently depleted by ET in the rainy season. Because very little water moved below the 200 cm depth, there was desiccation in the deep soil layer in drought and normal rainfall years.

Introduction

The Loess Plateau of China, located in the upper and middle reaches of the Yellow river, covers a total area of 62.85 × 10⁴ km² with an elevation of 1200–1600 m above sea level, and has a loess cover largely ranging from 30 to 80 m in thickness (Zhu et al., 1983, Shangguan and Zheng, 2006). This region is well-known for its serious soil erosion largely caused by cultivation of marginal lands and destruction of natural vegetation (Shi and Shao, 2000, Lu and van Ittersum, 2004, Wang et al., 2006). In order to control serious soil erosion, a great effort has been made to plant trees and grass on slope land since the end of 1950s. Indeed, about 24% of erosion area has been controlled, and vegetation coverage has increased from 6.5% in 1970s to 11% in 1995 in the Loess Plateau (He et al., 2003). However, large-scale vegetation restoration also aggravated water scarcity, and gradually led to soil desiccation in the deep soil layer (Li, 2001, Shangguan and Zheng, 2006, Chen et al., 2008). This desiccation conversely reduced the anti-drought capability of deep-rooted plants, and heavily influenced the growth and natural succession of vegetation. Artificial forest and grass degraded resulting in low yield and efficiency, and “small aged tree” with a height of 3–5 m appeared widely (Hou et al., 1999).

Soil desiccation often takes place below the depth of soil affected by rainfall infiltration (about 200 cm), and has relatively small and stable water content, which is between moisture of the capillary bond disruption and wilting point (Li, 2001). This desiccation usually results from the excessive depletion of deep soil water by artificial plants without sufficient rainwater supply. Soil desiccation was deserved little attention until 1980s, when it occurred widely in artificial forestland and grassland with a depth ranging from 100–300 to 800–1000 cm (Li, 2001, Chen et al., 2008). This desiccation has the horizontal and vertical distribution characteristics in the Loess Plateau. From southeast to northwest, soil desiccation becomes much heavier with lower water content and bigger range in depth due to drier climate and lower water holding capacity (Chen et al., 2008). However, this desiccation has some difference in range and intensity within the same hilly-gully region (Yang et al., 1999). It is often more intensive in forestland with a higher coverage, in forestland and grassland than that in cropland, and in vegetation-covered land than that in bare land. Furthermore, soil desiccation is usually more intensive on steep slope land than that on gentle slope land, and in hill and ridge than that in tableland or gully bottom.

From early Pleistocene to late Pleistocene, the development of loess deposit indicated the general trend of climate drought in north China, and loess was an indicator of the arid and semi-arid climate (Liu et al., 1985, Yang et al., 1999). According to the statistical data of drought and waterlogging from 1470 to 1979 in the Loess Plateau, the continuous drought with an average duration of 3.5 years took place 31 times and lasted 107 years, which covered by more than 75% of the total drought years (Yang et al., 1999). In most regions of the Loess Plateau, the average annual rainfall ranges from 300 mm in the northwest to 650 mm in the southeast, but the relevant mean annual evaporation varies from 623.8 to 1254.0 mm. This climate drought and loess soil properties suggested that soil desiccation might be a kind of physical phenomenon in semi-humid and semi-arid regions of the Loess Plateau (Yang et al., 1999, Chen et al., 2008). In fact, significant deep soil water uptake not only took place during the dry season, but also occurred during the wet season when rainfall was below average in most regions of the world (Jipp et al., 1998, Cameron, 2001, Oliveira et al., 2005, Seyfried et al., 2005, Sarris et al., 2007). However, when soil desiccation took place in the Loess Plateau, the insufficiency of deep soil water was difficult to disappear even if land use changed. For example, Hou et al. (1999) discovered that it would take more than 20 years to restore 0–350 cm water content in the degraded erect milkvetch (Astragalus adsurgens Pall.) grassland to that in natural grassland in Wuqi County of north Shaanxi. Mu et al. (2003) found that it would take about 15 years to resume 0–500 cm water content in artificial forestland to local natural conditions in Guyuan region of Ningxia. Huang and Gallichand (2006) using the SHAW model observed that the recovery time of soil water content varied from 6.5 to 19.5 years (average 13.7 years) in the 0–1000 cm soil profile, and from 4.4 to 8.4 years (average 7.3 years) in the upper 0–300 cm soil profile, after a 30-year-old apple (Malus pumila Mill.) orchard was converted to winter wheat (Triticum aestivum L.) in Changwu County of Shaanxi.

Soil desiccation in the deep soil layer impacted the water cycle in soil-plant-atmosphere systems by cutting off water supplies to groundwater, and reducing the total runoff into the Yellow river (He et al., 2003, Huang et al., 2003, Shangguan and Zheng, 2006). During the process of rainfall infiltration, one part of the infiltrated rainwater driven by the tension gradient in loess profile can move only to a limited range and cannot effectively supplement groundwater. The other part driven by gravity can move longer distance but soil water must reach the maximum field water holding capacity, and it is difficult to penetrate through the arid soil layer with a large amount of water deficit (Li, 2001, Shangguan and Zheng, 2006). If proper plant types were selected based on rainfall and soil water conditions, the negative effects of soil desiccation could be effectively minimized (Yang et al., 1999, Chen et al., 2008). However, even if the natural conditions of climate, hydrology and soil were taken into account, vegetation restoration would still have some significant effects on climate and the eco-hydrological environment (Fu and Yuan, 2001, Fu et al., 2003). This probably influenced soil water dynamics (Huszár et al., 1999, Rodriguez-Iturbe, 2000). Due to the lack of long-term in situ observations, the dynamic mechanism for water consumption of vegetation related to soil desiccation is still unknown. Therefore, there is a need to perform further physical experiments on soil water dynamics in various hydrological years in order to rehabilitate vegetation successfully.

The objectives of this study were to analyze the dynamic change in soil water with depth and the characteristics of water consumption and rainfall infiltration in soil, to understand soil water balance in grassland under natural and simulated rainfall conditions, and to further investigate the dynamic mechanism linked to desiccation that occurs deep in the soil profiles of the Loess Plateau.