The Effect of Robinia pseudoacacia Plantation on Soil Desiccation across Different Precipitation Zones of the Loess Plateau, China

: Ecological restoration has increased vegetation cover and reduced soil erosion, but it has also resulted in decreased soil-moisture content (SMC) and increased soil desiccation, which has ultimately led to a weakening of the “soil reservoir” function and a decline in the growth of plantations. Thus, soil desiccation has been a serious threat to the sustainable utilization of soil water resources and vegetation rehabilitation. In this study, the soil moisture of a Robinia pseudoacacia forest as well as its corresponding soil desiccation to a depth of 500 cm were measured across three different precipitation zones (400–450, 500–550 and 550–600 mm) along a north–south transect on the Loess Plateau. The results showed that the soil-moisture environment and soil desiccation status generally improved with the increasing precipitation gradient, while soil-moisture over-consumption signiﬁcantly declined ( p < 0.05). However, due to the elder forest-stand age and severe growth recession, the soil desiccation of R. pseudoacacia in the northern part was less than that in central zones. As the forest-stand age increased, SMC of R. pseudoacacia increased ﬁrstly and then decreased, and both soil-moisture consumption and the average soil desiccation rate peaked in the RP-5a, showing no signiﬁcant consistence with forest-stand age. Therefore, understanding the soil-moisture status of forestland may better provide scientiﬁc basis for native vegetation restoration and reconstruction in water-limited regions.


Introduction
Soil desiccation on the Loess Plateau is a unique phenomenon of soil hydrological deficit, which is formed by the joint action of various biotic and abiotic factors, such as vegetation, arid climate and aeolian soil. It is caused by an imbalance of the relationships between natural precipitation, soil-moisture storage and vegetation moisture consumption. Ultimately, soil desiccation may result in the significant reduction of deepsoil-moisture storage, as well as the formation of desiccated-soil layers located beneath the rainfall-infiltration layer, with lower soil-moisture content and relative persistence [1][2][3]. In addition, soil desiccation may bring about a series of ecological problems, such as a weakened "soil-reservoir" function, soil degradation, truncation between surface water and groundwater, deepened desiccated-soil layers as well as vegetation recession and death, and local climate drought [4,5]. Soil desiccation is widely reported at home and abroad, including in Russia [6], eastern Amazonia [7], southern Australia [8], the southwestern United States [9] and the Chinese Loess Plateau [5]. Since it was found in the southern edge of the Loess Plateau in 1960s, soil desiccation has gradually expanded to the whole plateau, and has increased desiccated-soil layers along with lowering the soil-moisture Notes: For the convenience and consistency of description, RP-Xa was used to illustrate the foreststand age of the sample plot. For example, RP-5a refers to the Robinia pseudoacacia plantation with 5-year stand age.

Field Sampling Measurements
In the northern and central regions of the Loess Plateau, a paired experiment design with R. pseudoacacia sampling sites and adjacent natural grassland was selected along In the typical northern, temperate, semi-arid, drought-prone steppe zone, the average annual air temperature was between 6.1 • C and 7.3 • C, the effective accumulated temperature above 10 • C was 2000-3200 • C, and the annual precipitation and potential evapotranspiration ranged from 400 to 500 mm and 1700 to 2500 mm, respectively. Additionally, the zonal soil consisted of sandy Aridisols with light to sandy loam according to Forests 2022, 13, 321 4 of 20 the USDA classification system. In the central, warm-temperate, semi-arid steppe zone, the average annual air temperature ranged from 9.5 • C to 10.6 • C and the effective accumulated temperature above 10 • C was between 2000 • C and 3000 • C, with the average annual precipitation and potential evapotranspiration between 500 and 550 mm and 1600 and 2100 mm, respectively. Additionally, the zonal soil transitioned from Aridisols to Alfisols with light loam to medium loam. In the southern, warm-temperate, semi-humid forest-steppe zone, the average annual air temperature and effective accumulated temperature above 10 • C were 9.0-9.2 • C and 3000-3500 • C, respectively. The average annual precipitation ranged from 550 to 600 mm and the mean annual potential evapotranspiration was between 1400 and 1900 mm [21,22,34]. Moreover, the zonal soil was clayed Alfisols with medium loam. Detailed background information regarding all of the sampling sites was summarized in Table 1.

Field Sampling Measurements
In the northern and central regions of the Loess Plateau, a paired experiment design with R. pseudoacacia sampling sites and adjacent natural grassland was selected along north-south direction [23]. In consideration of moisture-consumption depth, soil samples were collected at a depth of 500 cm in R. pseudoacacia and 300 cm in the natural grassland, respectively, with soil drilling (5 cm diameter) at an interval of 20 cm. At each sampling site, soil samples were collected with three duplicates from the same depth. To avoid the loss of soil moisture, soil samples were kept in sealed aluminum boxes in the field and then weighted in time. Simultaneously, sample sites were marked with GPS receivers, and plant species as well as geomorphologic characteristics such as altitude, slope gradient, slope aspect, slope position and other relevant information were also recorded. In addition, soil profiles with depths of 40 cm were excavated at each sampling site, and undisturbed soil was obtained using a cutting ring at the depths of 20 cm and 40 cm so as to determine the bulk-soil density and other physical parameters.

Soil-Desiccation Evaluation and Statistical Methods
Soil-moisture content (SMC) was measured by the oven-drying method at 105 • C for 24 h to obtain a constant mass. Gravimetric SMC was calculated by the soil-moisture loss from wet soil to drought soil during oven drying.
where SMC is the gravimetric soil-moisture content (%), G is the weight of the empty aluminum box (g), G 1 is the weight of the empty aluminum box and the wet soil before oven drying (g), and G 2 is the weight of the aluminum box with dried soil after oven drying (g). Soil-moisture storage (SMS) (mm) was calculated as follows: where SMS is the soil-moisture storage among all the soil profiles (mm), BD is the bulk-soil density (g/cm 3 ), SMC i is the gravimetric soil-moisture content at the ith depth of the soil profiles (%), and H i is the soil depth of the ith soil layer (mm). Available soil-moisture storage (ASMS) was represented as the difference between soil-moisture storage and that at the wilting soil moisture and was calculated as follows: where ASMS is the available soil-moisture storage among all the soil profiles (mm), SMS WM is the soil-moisture storage among all the soil profiles at the wilting soil moisture (mm).
Relative soil moisture (RSM) was used to evaluate the availability of soil moisture to the plant and was calculated as follows: where RSM is the relative soil moisture (%) and FC is the soil field capacity of each sampling site (%). Soil-desiccation index was calculated as follows: where SDI is the soil-desiccation index (%), WM is the wilting soil moisture (%), and SSM is the stable soil moisture of the soil-profile layers (%). Simultaneously, it was determined that a smaller SDI value represents a stronger intensity of soil desiccation and a greater SDI value represents a lower soil-desiccation intensity. On the basis of the above-calculated SDI, soil-desiccation intensity on the Loess Plateau was divided into six grades (Table 2). In general, stable soil moisture was regarded as the upper limit in evaluating soil desiccation under different vegetation types [21,22]. In the semi-humid and semi-arid areas of the Loess Plateau, on account of the constant variation in soil-moisture content along with annual and seasonal precipitation, as well as the difficulty in directly determining the stable deep-soil moisture in the short term, the SSM (stable soil moisture), which is the arithmetic mean value of wilting soil moisture and soil field capacity, was used to evaluate soil desiccation. Additionally, the SSM was about 50% to 70% of the soil-field capacity [21,22,35].

Soil-Moisture Content of R. pseudoacacia at Different Precipitation Zones
In the typical northern, temperate, semi-arid, drought-prone steppe zone (Figure 2a), the average SMC was between 1.97% and 13.14%, and the SMS varied from 84.42 to 834.26 mm (Table 3). Apart from the RP-30a of the Zhifanggou catchment, the SMC and SMS were significantly lower than that in the natural grassland, suggesting that at the same forest-stand age or in similar environmental conditions, the soil moisture of R. pseudoacacia in Ansai under good precipitation conditions was much higher than that in the other sampling sites. In addition to the RP-30a of Ansai County, the SMC in each sampling site was significantly lower than the stable soil moisture (11.35%). According to the stable soil moisture, all of the sampling sites exhibited varying degrees of soil desiccation except for the Zhifanggou catchment. By comparing the five sampling sites (Table 3), the ASMS was −138.85, −172.86, 62.25, 561.21 and 92.99 mm in turn from north to south, which was highly consistent with the aforementioned results. However, soil desiccation in the Hequ and Shenmu sites were the most severe with an average annual soil-desiccation rate of 11.58 mm per year and 11.48 mm per year, respectively.   Figure 2 were represented as the three different precipitation zones, namely, the typical northern, temperate, semi-arid, drought-prone steppe zone, the central, warm-temperate, semi-arid steppe zone, and the southern, warm-temperate, semi-humid forest-steppe zone, respectively. Arabic numerals indicate the forest-stand age of R. pseudoacacia plantation at each sampling site.
In the southern, warm-temperate, semi-humid forest-steppe zone (Figure 2c), the average SMC ranged from 7.49% to 21.03%, and the SMS varied from 486.72 mm to 1384.24 mm ( Table 3). The SMCs of the RP-15a and RP-20a at Renjiatai were lower than the stable soil moisture (13.09%) and the natural-grassland control (19.19%), leading to different degrees of soil desiccation. Meanwhile, their corresponding ASMS and soil-moisture overconsumption were 450.06 mm, 215.02 mm and 129.09 mm, 364.13 mm, respectively, and their average annual soil-desiccation rate reached 8.61 mm and 18.21 mm per year, which together indicated that soil-desiccation rate and soil-desiccation intensity gradually strengthened as the forest-stand age increased. However, the soil-moisture conditions of   Figure 2 were represented as the three different precipitation zones, namely, the typical northern, temperate, semi-arid, drought-prone steppe zone, the central, warm-temperate, semiarid steppe zone, and the southern, warm-temperate, semi-humid forest-steppe zone, respectively. Arabic numerals indicate the forest-stand age of R. pseudoacacia plantation at each sampling site. In the central, warm-temperate, semi-arid steppe zone (Figure 2b), the average SMC and SMS of each sampling site varied from 6.07% to 13.80%, and 435.50 mm to 897.26 mm, respectively (Table 3), where the SMC was the lowest for the RP-5a and highest for the RP-45a in the Yangjuangou catchment of Yan'an, but both of them were obviously lower than the corresponding natural grassland (15.44%). Apart from the RP-45a, the SMC of the remainder of the different-aged R. pseudoacacia in the Yangjuangou catchment were significantly lower than the stable soil moisture (13.25%) and exhibited diverse degrees of soil desiccation. However, the soil moisture of the RP-10a in the Angou catchment of Yanchang reached the second-lowest level with a mean SMC of 8.01% and an SMS of 520.65 mm, which were lower than the corresponding stable soil moisture (12.43%) but higher than the natural-grassland control (7.11%), and they also exhibited the phenomenon of soil desiccation. In addition, the ASMS of each sampling site ranged from 140.40 mm to 604.76 mm, and in which the RP-10a in Yanchang was the lowest, while the RP-45a in the Yan'an experimental area reached the highest. In view of the high soil-moisture over-consumption and the average annual soil-desiccation rate of R. pseudoacacia forests, the results further demonstrated the above conclusion that each sampling site exhibited different levels of soil desiccation, except for the RP-45a in Yangjuangou catchment which was of the optimal soil-moisture conditions.
In the southern, warm-temperate, semi-humid forest-steppe zone (Figure 2c), the average SMC ranged from 7.49% to 21.03%, and the SMS varied from 486.72 mm to 1384.24 mm ( Table 3). The SMCs of the RP-15a and RP-20a at Renjiatai were lower than the stable soil moisture (13.09%) and the natural-grassland control (19.19%), leading to different degrees of soil desiccation. Meanwhile, their corresponding ASMS and soilmoisture over-consumption were 450.06 mm, 215.02 mm and 129.09 mm, 364.13 mm, respectively, and their average annual soil-desiccation rate reached 8.61 mm and 18.21 mm per year, which together indicated that soil-desiccation rate and soil-desiccation intensity gradually strengthened as the forest-stand age increased. However, the soil-moisture conditions of the RP-10a and RP-15a at Zhaojiayuan were much better than that in Fuxian, and there was no soil desiccation or soil-moisture over-consumption, indicating a relatively high soil-moisture-conservation effect.
Generally, through comparing the soil-moisture conditions of each R. pseudoacacia sampling site across the three different precipitation zones, it could be concluded that the SMC, SMS and ASMS of each R. pseudoacacia sampling site showed increasing tendency, while the soil-moisture over-consumption significantly decreased as precipitation increased along the north-south direction. R. pseudoacacia plantations in the northern and central zones were found to have various degrees of soil desiccation, and the mean annual soildesiccation rate in the typical northern, temperate, semi-arid, drought-prone steppe zone was much smaller than that in the central, warm-temperate, semi-arid steppe zone.
In total, by comparing the soil-moisture conditions of different-aged R. pseudoacacia at the three different precipitation zones (Table 4), the results indicated that on account of the high degree of precipitation replenishment and low vegetation density in the RP-15a and the RP-45a, respectively, the SMC and ASMS in southern zone were much higher, whereas the SMC in the remainder of the sampling sites increased first and then decreased with the increase in forest-stand age. Soil-moisture over-consumption and the average annual soil-desiccation rate showed no apparent variation rules with forest-stand age, but both of them reached their highest level in the RP-5a.

Soil-Moisture Availability of R. pseudoacacia
The soil-moisture availability synthetically reflects the soil-moisture supply to the plant as well as the ease or complexity of soil-moisture utilization by plant. In this study, the relative soil moisture (RSM), which is the ratio of SMC to field capacity (FC), was used to evaluate the availability of soil moisture to vegetation growth according to the soilmoisture-availability classification results within different soil-texture zones by Yang on the Loess Plateau [36]. In view of the effects of different soil textures along the north-south direction on soil-moisture physical properties, the RSM range corresponding to each soilmoisture-availability grade varied slightly, of which the gravitational-infiltration aquifer and high-efficiency aquifer within different soil textures were consistent with each RSM range of greater than 100% and between 80% and 100%, respectively. The RSM range of the high-efficiency aquifer varied from 60% to 80% in sandy loam and medium loam, and from 50% to 80% in light loam, while the RSM range of mid-efficiency aquifer separately ranged from 25% to 59% in sandy loam, 30% to 49% in light loam and 35% to 59% in medium loam. Ultimately, the rest of the RSM ranges lower than the mid-efficiency aquifer were the invalid/low-efficiency aquifers.
As shown in Table 5, the SMC of R. pseudoacacia plantations generally increased as precipitation increased along the north-south direction. The proportion of invalid/lowefficiency aquifers and mid-efficiency aquifers accounting for the total soil profile gradually declined, while the proportion of high-efficiency aquifers and very-high-efficiency aquifers as well as their corresponding RSMs slowly increased. Even in the southern, warm-temperate, semi-humid forest-steppe zone, on account of the abundant rainfall replenishment, there appeared to be a relatively high proportion of gravitational-infiltration aquifers in the RP-15a forests of the Renjiatai and Zhaojiayuan sampling sites. In the typical northern, temperate, semi-arid, drought-prone steppe zone, the veryhigh-efficiency aquifer only appeared in the RP-30a of the Ansai sampling site with a proportion of 44%. The proportion of mid-efficiency aquifers and low-efficiency/invalid aquifers to the total profile in Hequ, Shenmu and Suide experimental areas were 16%, 6.25%, 60% and 84%, 93.75%, 40%, respectively, and the corresponding average RSMs were 28.59%, 31.43%, 34.09% and 20.57%, 14.06%, 27.91%, respectively. While, in the Ansai sampling sites, the high-efficiency aquifer in the RP-30a accounted for 56% of the total soil profile with an average RSM of 62.13%, and the high-efficiency aquifers, mid-efficiency aquifers and low-efficiency/invalid aquifers in the RP-40a separately accounted for 6.25%, 68.75% and 25% with the corresponding average RSMs of 51.63%, 39.43% and 35.24%, respectively.
In the central, warm-temperate, semi-arid steppe zone, apart from the RP-15a of the Yangjuangou catchment, there were no very-high-efficiency aquifers in the sampling sites. High-efficiency aquifers and mid-efficiency aquifers separately accounted for 4% and 96% with each homologous average RSM of 68.18% and 41.07% in the Angou catchment. The average RSM of the high-efficiency aquifers, mid-efficiency aquifers and lowefficiency/invalid aquifers in the RP-5a and RP-30a of the Yangjuangou catchment were 57.73%, 33.48%, 25.15% and 52.42%, 37.52%, 27.56%, and each aquifer accounted for 4%, 48%, 48% and 24%, 44%, 32% of the total soil profile, respectively. There were no highefficiency aquifers in the RP-15a of the Yangjuangou sampling site, while the very-highefficiency aquifers, mid-efficiency aquifers and low-efficiency/invalid aquifers separately accounted for 8%, 88%, 4%, with respective average RSMs of 83.64%, 35.23%, 27.73%. While in the RP-45a, there were only high-efficiency aquifer with an average RSM of 62.75%, further indicating a better soil-moisture condition than the other sites.
In southern, warm-temperate, semi-humid forest-steppe zone, neither the very-highefficiency aquifer nor high-efficiency aquifer were present in the RP-15a of the Fuxian experimental sites, and mid-efficiency aquifers and low-efficiency/invalid aquifers separately accounted for 72% and 12%, with each average RSM of 40.28% and 33.94%. Whereas, in the RP-20a, there were no very-high-efficiency aquifers, and the average RSMs of highefficiency aquifers, mid-efficiency aquifers and low-efficiency/invalid aquifers were 55%, 38.51% and 30.52%, accounting for 4%, 44% and 56% of the total soil profile, respectively. There were no mid-efficiency aquifers or low-efficiency/invalid aquifers in the two sampling sites of Yijun. Very-high-efficiency aquifers and high-efficiency aquifers accounted for 18.75% and 81.25% with respective average RSMs of 84.44% and 73.04% in the RP-10a, and 20% and 16% in the RP-15a with average RSMs of 90.86% and 75.71%, respectively. Besides, gravitational-infiltration aquifers also appeared in the RP-15a forests of the Fuxian and Yijun sampling sites, separately accounting for 16% and 64%, with average RSMs of up to 108.75% and 111.13%, indicating that abundant precipitation may result in ascendant soil-moisture conditions for R. pseudoacacia growth in semi-humid forest-steppe zones.

Soil Desiccation in R. pseudoacacia Forestland at Different Precipitation Zones
By evaluating the soil-desiccation intensities and the range of desiccated-soil layers within the soil profile of R. pseudoacacia forests along the north-south transect, the results showed that apart from the southern two sampling sites of Yijun, the remainder of the R. pseudoacacia sites exhibited different degrees of soil desiccation without exception (Table 6). In general, it was concluded that the SDI and the desiccated-soil-layer thickness (DSLT) gradually decreased as precipitation increased. The SDI in the northern experimental areas of the Hequ and Shenmu sites was the most severe with extreme soil desiccation, while in addition to the RP-45a in Yan'an sampling site, the SDI in the remainder of the central and southern zones reached slight to serious soil desiccation.
For example, in the typical northern, temperate, semi-arid, drought-prone steppe zone, the average SDI ranged from −100.78% to 125.36%, and extreme desiccated-soil layers occurred in the Hequ and Shenmu sites with a desiccated-soil-layer thickness (DSLT) of 500 cm and 320 cm, respectively. The SDI of the Suide site was intense soil desiccation, and its DSLT reached 500 cm, of which intense desiccated-soil layers and serious desiccated-soil layers separately comprised 340 and 160 cm. However, in the Ansai sampling sites, the SDI of the RP-30a was 125.36%, and there was no soil desiccation except for slight desiccatedsoil layers in the surface profile, while as the forest-stand age increased, the DSLT in the RP-40a reached up to 320 cm with intense, serious and medium desiccated-soil layers of 100, 120 and 100 cm, respectively.
In the central, warm-temperate, semi-arid steppe zone, except for the RP-45a of the Yan'an area exhibiting no soil desiccation, the SDI in the remainder of the sites reached serious soil desiccation. The total thickness of the desiccated-soil layers in the RP-10a of Yanchang was up to 480 cm, with intense, serious, medium and slight desiccated-soil layers of 200, 220, 40 and 20 cm, respectively. In the Yan'an experimental areas, the SDI was generally declined at first and then strengthened with the increase in forest-stand age in the RP-5a, RP-15a and RP-30a forests. Soil desiccation in the RP-5a and RP-30a forests were mainly comprised of intense and serious desiccated-soil layers each with DSLTs of 240 cm, 240 cm and 180 cm, 120 cm, respectively. While in the RP-15a, the total thickness of the desiccated-soil layers reached 460 cm, which was mainly occupied by medium soil desiccation with a thickness of 380 cm. In the southern, warm-temperate, semi-humid forest-steppe zone, on account of the highest precipitation and optimal soil-moisture conditions, soil desiccation did not exist in the Yijun sampling sites; however, soil-desiccation intensities strengthened as forest-stand age increased in the Fuxian experimental areas. For example, it was slightly desiccated in the RP-15a and the total thickness of the desiccated-soil layers was 420 cm with intense and medium desiccated-soil layers of 260 and 160 cm, respectively. Whereas serious soil desiccation with the total DSLT of 500 cm was found in the RP-20a, and the intense, serious, medium and slight desiccated-soil layers reached up to 80, 380, 20 and 20 cm, respectively.

Distribution of Desiccated-Soil Layers and Soil-Moisture Recovery of R. pseudoacacia
According to the definition of soil desiccation, desiccated-soil layers have an SMC lower than that of the stable soil moisture [21,22,35]. Thus, in our study, stable soil moisture and wilting soil moisture (WM) were used as the upper and lower bounds of desiccated-soil layers. As shown in Figures 3-5 and Table 7, the desiccated-soil layers of each sampling site within their total soil profile were designated on the basis of SDI value. Overall, the SDI and DSLT of R. pseudoacacia forests showed a declining tendency as precipitation increased. The soil-desiccation intensities of R. pseudoacacia from north to south reached intense, medium and no desiccation with SDI values of 6.90%, 50.86% and 114.08%, respectively. 7). It was found that soil-desiccation intensities on the whole gradually strengthened as forest-stand age increased. Soil-desiccation intensities within each soil layer in Hequ and Shenmu reached extreme soil desiccation and reached intense soil desiccation in Suide. To begin with, there only appeared to be slight soil desiccation below the depth of 300 cm in the RP-30a of the Ansai site, while soil-desiccation intensities strengthened to medium or above in the RP-40a with intense, serious and medium desiccated-soil layers at the interval depth of 100 cm, respectively. In the central, warm-temperate, semi-arid steppe zone, apart from slight desiccatedsoil layers at the surface of the soil profile in the RP-45a, there appeared to be slight or above desiccated-soil layers in the remainder of the sampling sites ( Figure 4 and Table 7). Soil-desiccation intensities reached medium or above in the RP-10a of the Yanchang site, at which it was intensely desiccated at a depth of 0-300 cm, and seriously and moderately desiccated at the 300-400 and 400-500 cm soil layers, respectively. Whereas, in the Yan'an experimental areas, soil-desiccation intensities alleviated at first and then strengthened with the increase in forest-stand age. Intense and serious desiccated-soil layers were present from the surface to 300 cm and 300 to 500 cm soil layers in the RP-5a, respectively. Slight and serious desiccated-soil layers were found in the RP-15a in the range of 0-100 and 100-500 cm soil layers, respectively. Additionally, in the RP-30a, serious desiccatedsoil layers were distributed in 0-200 and 300-400 cm ranges, and soil layers ranging between 200-300 and 400-500 cm were slightly and intensely desiccated, respectively.    In the southern, warm-temperate, semi-humid forest-steppe zone, there was no soil desiccation that appeared in the Yijun sampling site. However, soil-desiccation intensities gradually strengthened as forest-stand age increased in the RP-15a and the RP-20a of the Fuxian experimental areas ( Figure 5 and Table 7). For instance, there were only small amounts of serious and medium desiccated-soil layers in the distribution ranges of 100-400 and 400-500 cm in the RP-15a, respectively. In the RP-20a, medium desiccated-soil layers were mainly distributed from 0-100 cm, and soil layers below 100 cm were seriously desiccated.  Figure 4 were represented as sampling sites in Yanchang and Yan'an, respectively. desiccation that appeared in the Yijun sampling site. However, soil-desiccation intensities gradually strengthened as forest-stand age increased in the RP-15a and the RP-20a of the Fuxian experimental areas ( Figure 5 and Table 7). For instance, there were only small amounts of serious and medium desiccated-soil layers in the distribution ranges of 100-400 and 400-500 cm in the RP-15a, respectively. In the RP-20a, medium desiccated-soil layers were mainly distributed from 0-100 cm, and soil layers below 100 cm were seriously desiccated. On the vast Loess Plateau, atmospheric precipitation is the only replenishment to soil moisture. In order to investigate the required water amount and time for desiccated-soillayer recovery in R. pseudoacacia forests in different precipitation zones, the mean annual precipitation was calculated on the basis of integrated rainfall data from the past 60 years, and three different precipitation years, namely rainy year, normal year and dry year, were also divided based on the mean annual precipitation of ±10% [37]. Meanwhile, as shown in Table 8, the total amount of required water for soil-moisture recovery to local stable-  In the typical northern, temperate, semi-arid, drought-prone steppe zone, there appeared to be different degrees of soil desiccation and desiccated-soil layers in the soil profiles of R. pseudoacacia forests and even in the natural-grassland control (Figure 3 and Table 7). It was found that soil-desiccation intensities on the whole gradually strengthened as forest-stand age increased. Soil-desiccation intensities within each soil layer in Hequ and Shenmu reached extreme soil desiccation and reached intense soil desiccation in Suide. To begin with, there only appeared to be slight soil desiccation below the depth of 300 cm in the RP-30a of the Ansai site, while soil-desiccation intensities strengthened to medium or above in the RP-40a with intense, serious and medium desiccated-soil layers at the interval depth of 100 cm, respectively.
In the central, warm-temperate, semi-arid steppe zone, apart from slight desiccatedsoil layers at the surface of the soil profile in the RP-45a, there appeared to be slight or above desiccated-soil layers in the remainder of the sampling sites ( Figure 4 and Table 7). Soil-desiccation intensities reached medium or above in the RP-10a of the Yanchang site, at which it was intensely desiccated at a depth of 0-300 cm, and seriously and moderately desiccated at the 300-400 and 400-500 cm soil layers, respectively. Whereas, in the Yan'an experimental areas, soil-desiccation intensities alleviated at first and then strengthened with the increase in forest-stand age. Intense and serious desiccated-soil layers were present from the surface to 300 cm and 300 to 500 cm soil layers in the RP-5a, respectively. Slight and serious desiccated-soil layers were found in the RP-15a in the range of 0-100 and 100-500 cm soil layers, respectively. Additionally, in the RP-30a, serious desiccated-soil layers were distributed in 0-200 and 300-400 cm ranges, and soil layers ranging between 200-300 and 400-500 cm were slightly and intensely desiccated, respectively.
In the southern, warm-temperate, semi-humid forest-steppe zone, there was no soil desiccation that appeared in the Yijun sampling site. However, soil-desiccation intensities gradually strengthened as forest-stand age increased in the RP-15a and the RP-20a of the Fuxian experimental areas ( Figure 5 and Table 7). For instance, there were only small amounts of serious and medium desiccated-soil layers in the distribution ranges of 100-400 and 400-500 cm in the RP-15a, respectively. In the RP-20a, medium desiccated-soil layers were mainly distributed from 0-100 cm, and soil layers below 100 cm were seriously desiccated.
On the vast Loess Plateau, atmospheric precipitation is the only replenishment to soil moisture. In order to investigate the required water amount and time for desiccated-soillayer recovery in R. pseudoacacia forests in different precipitation zones, the mean annual precipitation was calculated on the basis of integrated rainfall data from the past 60 years, and three different precipitation years, namely rainy year, normal year and dry year, were also divided based on the mean annual precipitation of ±10% [37]. Meanwhile, as shown in Table 8, the total amount of required water for soil-moisture recovery to local stable-soilmoisture levels was calculated according to the measured SMC and stable-soil-moisture content. It was found that the SMC in the RP-30a of Ansai, the RP-45a of Yan'an and the RP-15a of Yijun were, respectively, higher than their corresponding regional stable soil moistures; thus, the annual precipitation could satisfy the plant growth and the soil moisture could be fully restored. However, the recovery time for the other sampling sites was at least one year or more. In general, as precipitation decreased and forest-stand age increased along the south-north transect, both the soil-moisture-recovery difficulty and the required water amount and time gradually increased.
In the typical northern, temperate, semi-arid, drought-prone steppe zone, the annual precipitation ranged from 232.85 to 719.6 mm with an average annual precipitation of 434.00 mm from 1953 to 2014, and the average precipitation of rainy, normal and dry years was 477, 434 and 391 mm, respectively. However, the water demand for crops was about 300 mm per year [21], so the precipitation supply for soil-moisture recovery in each precipitation year was 177, 134 and 91 mm, respectively. Therefore, it could be concluded that the required time for soil-moisture recovery to stable-soil-moisture levels in the three continuous precipitation years (rainy, normal and dry precipitation years) were at least 3, 4 and 6 years in Hequ, 2, 3 and 4 years in Shenmu and Suide, and 3, 4 and 5 years in Ansai, respectively. In short, considering the three precipitation years, soil-moisture recovery to stable-soil-moisture levels needed more than 4, 3, 3 and 4 years, respectively. Similarly, in the central, warm-temperate, semi-arid steppe zone, the annual precipitation was 536.39 mm and the average precipitation in the rainy and dry years was 590 and 482 mm, respectively. Based on the local water demand for crop growth of 350 mm, the surplus precipitation for soil-moisture recovery in rainy, normal and dry years was 240, 186 and 132 mm, respectively [21]. Thus, it was determined that the required time for soil-moisture recovery to stable-soil-moisture levels in the RP-10a of Yanchang were 2, 2 and 3 years in continuous rainy, normal and dry years. In the Yan'an experimental areas, it needed 2, 2 and 3 years in the RP-5a, and 2, 2 and 3 years in the RP-15a and RP-30a, respectively. Taking each precipitation year into account, in general, it separately needed more than 2 years in the RP-10a of Yanchang and 3, 2 and 2 years in the RP-5a, RP-15a and RP-30a of Yan'an to recover desiccated-soil moisture.
In the southern, warm-temperate, semi-humid forest-steppe zone, the average precipitation in rainy, normal and dry years was 660, 600 and 540 mm, respectively, and the annual water demand for local crops was about 400 mm [21,22]. Thus, it could be calculated that the precipitation for desiccated-soil moisture recovery was 260, 200 and 140 mm, respectively. In continuous rainy, normal and dry years, in order to recover desiccated-soil moisture to stable-soil-moisture levels, it needed more than 1 year in the RP-15a of Fuxian, more than 2, 2 and 3 years in the RP-20a of Fuxian and at least 1 or 2 years in the RP-10a of Yijun, respectively. Overall, in consideration of the various precipitation years, it separately required more than 1 and 2 years in the RP-15a and RP-20a of Fuxian, and more than 1 year in the RP-10a of Yijun for desiccated-soil moisture recovery.

Soil-Moisture Conditions of R. pseudoacacia at Different Precipitation Zones
Due to its strong adaptability and fast growth, R. pseudoacacia turns out to be the main tree-planting species for the Reforestation Project and is widely distributed from the northern arid and semi-arid areas to southern semi-humid areas on the vast Chinese Loess Plateau [23]. However, as pointed out in previous studies, precipitation is the only source of soil-moisture supply owing to the deep watertable buried below the thick Loess soils on the Loess Plateau [38,39]. Therefore, on account of the differences among various precipitation types along the south-north direction, rainfall gradually decreases as dryness increases. And in the same direction, precipitation also shows a descending tendency with the increase in latitude and elevation [23]. Thus, in southern areas, rainfall was abundant and soil-moisture holding capacity as well as deep-soil-moisture storage were high. In the northern and central hilly region, deep-soil-moisture storage was always kept at lowhumidity conditions and was mostly below the stable-soil-moisture level, which conversely resulted in obvious regional differentiation for R. pseudoacacia growth [23,40]. By comparing the soil-moisture conditions of R. pseudoacacia forests at different precipitation zones, there appeared to be various degrees of soil-moisture deficit from the southern forest zone to the northern steppe zone [41]. Both of their SMC, SMS and ASMS values increased as precipitation increased along the north-south direction, whereas soil-moisture over-consumption significantly declined (Tables 3 and 4). Based on the score of the relatively greater foreststand age as well as the severe degradation, the average annual soil-desiccation rate in the northern zone was lower than that in central zone.
To be specific, in southern areas, owing to the combined effect of strong root-water uptake as well as rapid transpiration of water consumption as forest-stand age increased, the soil-moisture deficit was severely increased in Fuxian sampling sites [42]. Conversely, on account of the orographic effect of Huanglong and Ziwuling Mountains, there appeared to be more pluvial areas with abundant precipitation, which resulted in relatively high soil-moisture conditions in R. pseudoacacia forests. However, in northern and central areas, owing to little precipitation combined with excessive soil-moisture consumption by plant growth [23,40], the soil-moisture deficit was extremely severe, especially in the Hequ and Shenmu sampling sites, with each having an ASMS and average annual soil-desiccation rate of −138.85, −172.86 mm and 11.58, 11.48 mm per year, respectively ( Table 2).
In addition, as precipitation increased, soil-moisture availability improved significantly. The low-efficiency/invalid aquifers and mid-efficiency aquifers within the total soil profiles gradually decreased, while high-efficiency aquifers and very-high-efficiency aquifers as well as their corresponding average RSMs progressively increased. Even in the southern, warm-temperate, semi-humid forest-steppe zone, there existed a relatively high proportion of gravitational-infiltration aquifers in the RP-15a forests in both the Yijun and Fuxian experimental areas.
Existing studies have shown that the soil-moisture conditions of R. pseudoacacia forests on the Loess Plateau could be approximately classified into two groups based on the boundary of the Yan'an experimental areas [23,25]. The soil-moisture availability in the southern sampling sites such as Huangling and Yijun was represented as mid-efficiency and there was almost no soil-moisture deficit [41] or significant soil desiccation. To some extent, soil moisture could be satisfied by regular plant growth, and meanwhile, the shrublawn structure beneath the arbor stratum had also undergone a significant change [23], indicating that vegetation communities were in steady positive succession with improved structure and function, and continued to play key roles in maintaining ecological protection as well as soil and water conservation. However, the soil-moisture storage in most of the northern areas apparently declined, and their soil-moisture availabilities reached a low to medium efficiency level [41], leading to various degrees of soil-moisture deficit and persistent desiccated-soil layers below 160 cm beneath the soil surface. Our results were greatly consistent with Wang [25], who determined that the average SMC was close to the wilting soil moisture (WM) even in some sampling sites at the depths of 300 and 500 cm soil profiles [25], which ultimately constrained R. pseudoacacia growth and resulted in extensive degradation and defoliation in the normal growing seasons.

Soil Desiccation of Different-Aged R. pseudoacacia
Deep-soil desiccation on the Loess Plateau was caused by the accumulated soilmoisture consumption of forests and grasses. On the whole, soil-moisture consumption by transpiration will gradually increase as forest-stand age increases [40], and if it exceeds the soil-moisture holding capacity without timely adjustment, then soil desiccation will be ensure and soil-moisture conditions will further deteriorate [25]. Meanwhile, soil desiccation was closely associated with forest-stand age [1,35,36], and the DSLT showed a positive correlation with forest-stand age [12]. By comparing the SMC of different-aged R. pseudoacacia forests, it was found that the SMC linearly decreased at first, and then increased with fluctuation as forest-stand age increased, which was consistent with the regional SMC distribution [23]. Both soil-moisture over-consumption and the average annual soil-desiccation rate showed no significant relationships; however, they reached their highest level in the RP-5a in the central, warm-temperate and semi-arid steppe zone. Given the massive soil-moisture consumption at the fast-growing stage, together with greater vegetation density, soil moisture was extremely consumed [21,22], and thus led to the highest soil-moisture over-consumption and average annual soil-desiccation rates with values of 425.75 and 85.15 mm per year, respectively.
Specifically, the RP-40a forests were mainly distributed in the typical northern, temperate, semi-arid, drought-prone steppe zone, which saw scarce precipitation and intense evaporation, as well as strong root-water uptake, thereby leading to the lowest SMC and ASMS (4.61%, −24.67 mm, respectively). Conversely, the RP-20a forests were mainly distributed in Renjiatai sampling site of the southern, warm-temperate, semi-humid foreststeppe zone, and vegetation grew exuberantly with larger density, which together resulted in the second-highest soil-moisture over-consumption and average annual soil desiccation rates than the RP-10a and the RP-30a in the northern and central zones.
In general, as the forest-stand age increased, the older R. pseudoacacia could effectively improve soil quality, soil porosity as well as soil-moisture holding capacity with the interaction of forest litter and plant roots [43][44][45]. Besides, once the surface moisture was exhausted, R. pseudoacacia could also fully utilize the deep-soil moisture to satisfy its growth, which ultimately resulted in the further deterioration of soil moisture [46]. However, it was found that the SMC of the RP-45a reached the highest level in Yangjuangou catchment of the Yan'an sampling site, indicating that the aging forest vegetation could cause community density to decline by continuous self-thinning, leaving the vegetation community to maintain self-succession and sustainable development in order to retain more soil moisture [34]. Therefore, the soil-moisture conditions in the RP-45a was optimal.

Soil Desiccation and Desiccated-Soil Moisture Recovery of R. pseudoacacia
On the vast Chinese Loess Plateau, due to the specific hydrological characteristics, the rainfall-infiltration depth was generally less than 200 cm with no deep percolation [1,23,47]. Additionally, in view of the dry climate, plants necessarily absorbed soil moisture from the deeper soil layers through elongated roots to obtain replenishment. Therefore, it was difficult to be restored in a short amount of time once the deep-soil moisture was exhausted, which may have consequently led to a decreasing deep-soil-moisture storage and the formation of thick desiccated-soil layers. On the contrary, desiccated-soil layers under the R. pseudoacacia forest restricted its deep-soil-moisture absorption in drought years and limited its normal development as well, which finally caused widespread inefficient and low-yield artificial forests, and thus posed a great threat to vegetation restoration [25]. For example, at different precipitation zones, R. pseudoacacia forests in the northern, central and southern areas exhibited intense, medium and no desiccation, respectively. It was concluded that soil-desiccation intensities and the thickness of desiccated-soil layers gradually decreased with the increasing precipitation. In the northern and central sampling sites, soil desiccation of R. pseudoacacia was greatly related to both the dry climate and the location of sunny slopes. Meanwhile, as the forest-stand age and the degree of drought increased, the vertical distribution depth and the density of fine roots significantly increased, and soil-moisture consumption as well as water demand for plant transpiration gradually rose, which together led to the decline of soil moisture and the formation of intense and medium desiccated-soil layers [47]. Conversely, in southern R. pseudoacacia sampling sites, as precipitation significantly increased, soil-desiccation intensities gradually alleviated. Although slight soil desiccation was seen in some sampling sites, the soil moisture could be restored after precipitation replenishment during the rainy seasons and the annual precipitation could meet its normal growth.
Desiccated-soil moisture recovery and replenishment were closely related to rainfall amounts in each precipitation year on the plateau [1,2,48]. In general, soil moisture could be evenly compensated in most years through rainy seasons in the southern and southeastern areas, where precipitation was much more abundant. Conversely, as temperature and precipitation decreased northward and northwestward, soil moisture could only be slightly restored in the minority humid years [4], and soil moisture was always imbalanced in the areas with precipitation of less than 200 mm. At the three different precipitation zones, only in few sampling sites, such as the RP-30a of Ansai, the RP-45a of Yan'an and the RP-15a of Yijun, the SMCs were higher than the regional stable soil moisture, and soil moisture could be fully restored. However, the required time for soil-moisture recovery in the remainder of the sampling sites was more than 1 year, and as precipitation decreased and forest-stand age increased, the difficulty, water demand, and the required time for soil-moisture recovery gradually increased, with requirements of 3 or 4 years in the central and northern zones and at least 1 year in southern areas. Although precipitation gradually reduced from southeast to northwest on the whole of the Loess Plateau, there were more pluvial regions by the terrain effect of Huanglong and Ziwuling mountains in the southern areas, such as Yijun, thus the soil moisture could also be restored during the rainy seasons in most years [25].

Conclusions
By analyzing the measured soil-moisture profile and evaluating the soil-desiccation intensities of R. pseudoacacia at different precipitation zones along north-south direction, the conclusions were as follows.
As precipitation increased, the SMC, SMS and ASMS of R. pseudoacacia forests gradually increased, while soil-moisture over-consumption significantly declined. Due to the relatively greater forest-stand age, R. pseudoacacia in the northern zone were severely degraded, and their average annual soil-desiccation rate was less than that in central sampling sites. Additionally, the SMC increased at first and then decreased with the increase in foreststand age. Although soil-moisture over-consumption and average soil-desiccation rate showed no significant relationships with forest-stand age, both of them reached the highest level in the RP-5a.
Meanwhile, the proportion of low-efficiency/invalid and mid-efficiency aquifers to the total soil profiles gradually decreased, while the proportion of high-efficiency and very-highefficiency aquifers as well as their corresponding average RSMs slowly increased with the increase in precipitation. Even in the southern, warm-temperate, semi-humid forest-steppe zone, there also existed a relatively high proportion of gravitational-infiltration aquifers.
Besides, the soil-desiccation intensities of each R. pseudoacacia sampling site decreased from north to south as a whole, reaching intense, medium and no soil desiccation, respectively, and the thickness of desiccated-soil layers and the difficulty of soil-moisture recovery gradually reduced with the increased precipitation. However, in view of different precipitation years, at least two years or more were needed for the slow desiccated-soil moisture recovery to stable-soil-moisture levels.
Therefore, in combination with the soil-moisture conditions of each R. pseudoacacia sampling site and the precipitation amount along the north-south transect, the self-succession of natural vegetation should be implemented in the southern, warm and humid areas such as Yijun, where the annual precipitation can meet the vegetation growth requirements. Conversely, in the central and northern drought zones, as forest-stand age increases, selfthinning of R. pseudoacacia forests can gradually desiccated-soil moisture, but the process is relatively slow. Hence, nature-based recovery together with increasing artificial-catchment measures should be undertaken in order to restore soil moisture to stable-soil-moisture levels as soon as possible.