Root Distribution and Root Cohesion of Two Herbaceous Plants in the Loess Plateau of China

: In order to understand the root morphology distribution and mechanical properties of typical herbaceous plants, and to evaluate the ability of soil reinforcement by the plant roots, root morphology investigation, single root tensile test in laboratory and root cohesion evaluation by the Wu-Waldron model were carried out on two local representative herbaceous plants, Kochia scoparia (L.) Schrad and Artemisia sacrorum Ledeb. in the Loess Plateau of China. The results showed that the root morphological indexes (root number, single root diameter, root cross-sectional area, root surface area, root volume and root area ratio) of the two herbaceous plants decreased with the increase in soil depth, and the ratio of root to shallow soil layer was the highest in the 0–10 cm soil layer. The efﬁciency of root reinforcement could be higher in the shallow soil layer less than 10 cm. A positive correlation was observed between the root tensile force and root diameter in power function or exponential function, and a negative correlation was observed between the root tensile strength and root diameter in power function. The root cohesion of Kochia scoparia (2.73 kPa, or 0.92 kPa–1.37 kPa) was greater than that of Artemisia sacrorum (1.60 kPa, or 0.54 kPa–0.8 kPa), which could be used as the preferred herbaceous plant species for soil erosion control. The results could provide a scientiﬁc basis for selecting dominant species in the ﬁelds of ecological slope protection and soil and water conservation plant engineering in the loess area. soil layers and among the 15–30 cm soil layers. The roots of Artemisia sacrorum were distributed in the 0–40 cm soil layer. The maximum average root number was 45 in the 0–5 cm soil layer. average root number decreased with soil depth. The number of roots in the 0–15 cm soil layers was significantly greater than that in the 15–40 cm soil layers.


Introduction
The Loess Plateau is located in the arid and semi-arid area of China, having only 200-700 mm precipitation and half of that in the summer [1]. The soil of the Loess Plateau is loose with vertical joints, which is easily eroded. The soil erosion modulus in most areas ranges from 1000 t/km −2 a −1 to 5000 t/km −2 a −1 . The soil erosion modulus in Huangpuchuan Basin is the highest, which is close to 25,000 t/km −2 a −1 [2,3]. The loose and porous texture of loess, uneven distribution of precipitation and unreasonable land use by humans leads to frequent geological disasters, such as slope destabilization, which aggravates soil erosion. Vegetation is the most active and effective factor to prevent soil erosion. Since as early as the 1960s, the United States has attached great importance to the role of slope ecological engineering in soil erosion and slope destabilization [4]. In recent years, the use of plants to prevent soil erosion, slope instability and other geological disasters has also been widely acknowledged by scholars worldwide [5][6][7][8].
The mechanical effect of plants on soil reinforcement is mainly manifested in that plant roots can transfer soil shear stress through their own tension and improve soil shear strength. Wu et al. [9] show that plant roots could improve soil shear strength mainly by improving soil cohesion, and have no obvious effect on the soil's internal friction angle. Therefore, the mechanical effect of soil reinforcement by roots can be explained

Study Area
The roots of Kochia scoparia and Artemisia sacrorum were investigated and sampled from a hilly slope in the Western Hill Ecological Park, Taiyuan City, Shanxi Province, China (37 • 84 N, 112 • 46 E). The study area has a warm temperate continental monsoon climate with long dry and cold winters, hot and humid summers, sharp temperature rises in spring and rapid cooling in autumn. The solar radiation is strong with rich light energy and heat there and the annual sunshine time can reach 2400~2600 h. The temperature varies widely from day to night. The annual average temperature is 9.5 • C. The rainfall is mainly concentrated in July to September and the annual average precipitation is about 470 mm [31]. The soil here is sandy loam (Table 1).

Root Morphological Investigation
The root morphology and root tensile characteristics of the two herbaceous plants, Kochia scoparia and Artemisia sacrorum, which are widely distributed in the study area, were studied. Kochia scoparia is an annual herbaceous plant and Artemisia sacrorum is a perennial herbaceous plant. The root morphological distribution was investigated on Kochia scoparia (five individuals) and Artemisia sacrorum (five individuals) with the whole-plant excavation method described by [14,32]. Soil columns with roots were dug out completely and put into sealing bags. The radius of the columns was 20 cm for Kochia scoparia and 30 cm for Artemisia sacrorum determined by the maximum horizontal distribution of the plant roots. The excavation depth was 40 cm as the soil below the depth was hard and the roots of the herbs were usually distributed up to the depth. The bags of soil columns were taken back to the laboratory under −4 • C storage [11]. The roots were then cleaned carefully in the laboratory every 5 cm layer from the top to the bottom. The number of roots in each layer was counted and recorded. The length of the roots was measured with a steel ruler with an accuracy of 1 mm. The root diameter was measured by an electronic vernier caliper with an accuracy of 0.01 mm. The root diameter of the roots in each soil layer was measured three times. The average value was considered as the root diameter used for the calculation of root cross-sectional area. Kochia scoparia had obvious lateral roots in addition to the main roots (Figure 1a), while Artemisia sacrorum had primary roots without obvious lateral roots (Figure 1b). Therefore, the primary roots and lateral roots were measured for Kochia scoparia, while only the primary roots were measured for Artemisia sacrorum. After the measurement of root diameter and root length in each layer, the root indexes, including root cross-sectional area s, root surface area S, root volume V and root area ration (RAR) in each layer were calculated using the following equations [18,33]: where N is the number of roots in each layer, D is the diameter of a single root (mm 2 ), h is the soil thickness of each layer (mm), Ar is the cross-sectional area of the roots on the shear plane (mm 2 ) and As is the total area of the soil shear surface (mm 2 ). The projection area of the maximum circular root distribution range on the horizontal surface was equal to As in this study.

Root Tensile Tests
Root samples with good growth, undamaged appearance and complete diameter gradient were selected for the root tensile tests. A total of 68 root samples of Kochia scoparia and 32 root samples of Artemisia sacrorum were tested with the method by [14]. The WDW-5 electronic universal testing system (Changzhou Sanfeng Instrument Technology Co., Ltd., Changzhou, China) was used to determine the peak tensile force F (N) and tensile strength T R (MPa) of the single root of Kochia scoparia and Artemisia sacrorum. A gauge length of Sustainability 2022, 14, 8053 4 of 12 100 mm and strain rate of 100 mm·min −1 were set in the tests [34]. Before stretching, the diameter of the upper, middle and lower root was measured with an electronic vernier caliper. The average value was taken as the average root diameter of the root sample. The equipment automatically records the maximum tensile resistance of the root specimen when tensile failure occurs. In order to ensure the root activity, the collected root samples were measured within two days. The root tensile strength was calculated by the following formula [18]: where T R is root tensile strength (MPa), F is the maximum tensile force (N) and D is average root diameter (mm).

Root Tensile Tests
Root samples with good growth, undamaged appearance and complete diameter gradient were selected for the root tensile tests. A total of 68 root samples of Kochia scoparia and 32 root samples of Artemisia sacrorum were tested with the method by [14]. The WDW-5 electronic universal testing system (Changzhou Sanfeng Instrument Technology Co., Ltd., Changzhou, China) was used to determine the peak tensile force F (N) and tensile strength (MPa) of the single root of Kochia scoparia and Artemisia sacrorum. A gauge length of 100 mm and strain rate of 100 mm•min −1 were set in the tests [34]. Before stretching, the diameter of the upper, middle and lower root was measured with an electronic vernier caliper. The average value was taken as the average root diameter of the root sample. The equipment automatically records the maximum tensile resistance of the root specimen when tensile failure occurs. In order to ensure the root activity, the collected root samples were measured within two days. The root tensile strength was calculated by the following formula [18]: where is root tensile strength (MPa), F is the maximum tensile force (N) and D is average root diameter (mm).

Root Cohesion Evaluation
The Wu-Waldron model is widely used to evaluate the root cohesion of plants [9,35]. The root cohesion (Cr) in root-soil composites is calculated by the formula [33]: where is root tensile strength on soil shear surface (kPa) and is the root area ratio RAR. Based on the root tensile strength and root area ratio, the increased cohesion in the

Root Cohesion Evaluation
The Wu-Waldron model is widely used to evaluate the root cohesion of plants [9,35]. The root cohesion (Cr) in root-soil composites is calculated by the formula [33]: where T R is root tensile strength on soil shear surface (kPa) and A R A S is the root area ratio RAR. Based on the root tensile strength and root area ratio, the increased cohesion in the soil due to the presence of roots can be estimated. These two parameters are called the biotechnology root characteristic index [36].

Data Analysis
The data were analyzed using SPSS 20.0. Analysis of covariance (ANCOVA) was used to analyze the differences in the parameters of the root distribution under different soil depths. An independent-samples T-test was used to compare the difference in the mean values of the root parameters of the two plants. The significance was at the 0.05 level. Different lowercase letters indicate significant differences in the parameters between soil depths. The graphs were plotted using Excel 2013.

Root Number
The number of roots of the two herbaceous plants was related with soil depth in exponential functions (Kochia scoparia y = 154.66e − 0.564x, R 2 = 0.819; Artemisia sacrorum y = 136.34e − 0.561x, R 2 = 0.912; Figure 2). The roots of Kochia scoparia were mostly distributed in the 0-30 cm soil layer. The number of roots increased first and then decreased with the increase in soil depth. The maximum average root number was 59 in the 5-10 cm soil layer. No significant difference in the number of roots was observed among the 0-15 cm  Figure 2). The roots of Kochia scoparia were mostly distributed in the 0-30 cm soil layer. The number of roots increased first and then decreased with the increase in soil depth. The maximum average root number was 59 in the 5-10 cm soil layer. No significant difference in the number of roots was observed among the 0-15 cm soil layers and among the 15-30 cm soil layers. The roots of Artemisia sacrorum were distributed in the 0-40 cm soil layer. The maximum average root number was 45 in the 0-5 cm soil layer. The average root number decreased with soil depth. The number of roots in the 0-15 cm soil layers was significantly greater than that in the 15-40 cm soil layers.

Root Diameter
The average root diameter of Kochia scoparia and Artemisia sacrorum decreased in power functions with soil depth. (Kochia scoparia = 0.6687 −0.495 , 2 = 0.990; Artemisia sacrorum = 0.7403 −0.28 , 2 = 0.987; Figure 3). The root diameter of Kochia scoparia ranged from 0.06 mm to 2.69 mm with an average diameter of 0.69 ± 0.06 mm in the 0-5 cm soil layer. The root diameter in 0-10 cm soil layers was significantly larger than that in 10-30 cm soil layers. The diameter of Artemisia sacrorum ranged from 0.06 mm to 2.19 mm. The maximum average root diameter was observed in 0-5 cm soil layer (0.75 ± 0.08 mm), which was significantly larger than that in 10-40 cm soil layers (0.47 ± 0.05 mm).

Other Root Morphological Indexes
The root cross-sectional area, root surface area, root volume and root area ratio of the two herbaceous plants all decreased exponentially with the increase in soil depth ( Table 2). The maximum root cross-sectional areas of Kochia scoparia and Artemisia sacrorum were 19.73 mm 2 and 21.67 mm 2 observed in the 0-5 cm soil layer, which was significantly greater than that in 10-40 cm soil layers (Figure 4a).

Root Diameter
The average root diameter of Kochia scoparia and Artemisia sacrorum decreased in power functions with soil depth. (Kochia scoparia y = 0.6687x −0.495 , R 2 = 0.990; Artemisia sacrorum y = 0.7403x −0.28 , R 2 = 0.987; Figure 3). The root diameter of Kochia scoparia ranged from 0.06 mm to 2.69 mm with an average diameter of 0.69 ± 0.06 mm in the 0-5 cm soil layer. The root diameter in 0-10 cm soil layers was significantly larger than that in 10-30 cm soil layers. The diameter of Artemisia sacrorum ranged from 0.06 mm to 2.19 mm. The maximum average root diameter was observed in 0-5 cm soil layer (0.75 ± 0.08 mm), which was significantly larger than that in 10-40 cm soil layers (0.47 ± 0.05 mm).  The total root surface area of Kochia scoparia and Artemisia sacrorum was 14,827.30 mm 2 and 16,629.95 mm 2 , respectively. The root surface area in the 0-5 cm shallow soi layer accounted for 36.02% (Kochia scoparia) and 32.92% (Artemisia sacrorum) of the tota root surface area, and it was significantly greater than that in 25-40 cm soil layers ( Figure  4b).
The total root volume was 2023.00 mm 3 and 2717.26 mm 3 , respectively. In the 0-5 cm soil layer, root volume accounted for 48

Other Root Morphological Indexes
The root cross-sectional area, root surface area, root volume and root area ratio of the two herbaceous plants all decreased exponentially with the increase in soil depth ( Table 2). The maximum root cross-sectional areas of Kochia scoparia and Artemisia sacrorum were  19.73 mm 2 and 21.67 mm 2 observed in the 0-5 cm soil layer, which was significantly greater than that in 10-40 cm soil layers (Figure 4a).  The total root surface area of Kochia scoparia and Artemisia sacrorum was 14,827.30 mm 2 and 16,629.95 mm 2 , respectively. The root surface area in the 0-5 cm shallow soil layer accounted for 36.02% (Kochia scoparia) and 32.92% (Artemisia sacrorum) of the total root surface area, and it was significantly greater than that in 25-40 cm soil layers ( Figure  4b).
The total root volume was 2023.00 mm 3 and 2717.26 mm 3 , respectively. In the 0-5 cm soil layer, root volume accounted for 48.77% (Kochia scoparia) and 39.88% (Artemisia sacrorum) of the total root volume. For Kochia scoparia, the root volume in the 0-5 cm soil layer was significantly higher than that in 5-30 cm soil layers. For Artemisia sacrorum, the root volume in the 0-15 cm soil layers was significantly higher than that in 15-40 cm soil layers (Figure 4c).
The root area ratio (RAR) reflects the proportion of root area in the soil, and it has an important effect on soil physical properties and the absorption of soil nutrients by roots. The RAR of Kochia scoparia was 0.0365% in the 0-5 cm soil layer and then decreased to 0.0004% in the 25-30 cm soil layer. The RAR of Artemisia sacrorum was 0.0272% in the 0-5 cm soil layer, and then decreased to only 0.0002% in the 35-40 cm soil layer. The RAR was significant greater in the 0-10 cm soil layers than in the soil layers below 10 cm (Figure 4d).     The total root surface area of Kochia scoparia and Artemisia sacrorum was 14,827.30 mm 2 and 16,629.95 mm 2 , respectively. The root surface area in the 0-5 cm shallow soil layer accounted for 36.02% (Kochia scoparia) and 32.92% (Artemisia sacrorum) of the total root surface area, and it was significantly greater than that in 25-40 cm soil layers (Figure 4b). The total root volume was 2023.00 mm 3 and 2717.26 mm 3 , respectively. In the 0-5 cm soil layer, root volume accounted for 48.77% (Kochia scoparia) and 39.88% (Artemisia sacrorum) of the total root volume. For Kochia scoparia, the root volume in the 0-5 cm soil layer was significantly higher than that in 5-30 cm soil layers. For Artemisia sacrorum, the root volume in the 0-15 cm soil layers was significantly higher than that in 15-40 cm soil layers (Figure 4c).
The root area ratio (RAR) reflects the proportion of root area in the soil, and it has an important effect on soil physical properties and the absorption of soil nutrients by roots. The RAR of Kochia scoparia was 0.0365% in the 0-5 cm soil layer and then decreased to 0.0004% in the 25-30 cm soil layer. The RAR of Artemisia sacrorum was 0.0272% in the 0-5 cm soil layer, and then decreased to only 0.0002% in the 35-40 cm soil layer. The RAR was significant greater in the 0-10 cm soil layers than in the soil layers below 10 cm (Figure 4d).

Root Tensile Properties
In the root tensile tests, 33

Root Cohesion
The root cohesion of the two herbaceous plants decreased in logarithmic functions with soil depth (Kochia scoparia = −5.243 ln + 18.084, 2 = 0.971; Artemisia sacrorum = −6.81 ln + 19.595, 2 = 0.953; Figure 6). The maximum values appeared in the soil depth of 0-5 cm, in which the cohesion of Kochia scoparia was 8.04 kPa, while that of Artemisia sacrorum was 5.18 kPa. There was no difference in the cohesion between the two plants in different soil layers below a 10 cm soil depth. The cohesion of the 0-10 cm soil layer was significantly different from that of the 10-30 cm soil layer, and that of the 0-10 cm soil layer was significantly higher than that of the 10-40 cm soil layer. The average root cohesion of the two plants was Kochia scoparia (2.73 kPa) > Artemisia sacrorum ( Root cohesion (kPa)

Root Cohesion
The root cohesion of the two herbaceous plants decreased in logarithmic functions with soil depth (Kochia scoparia y = −5.243 ln x + 18.084, R 2 = 0.971; Artemisia sacrorum y = −6.81 ln x + 19.595, R 2 = 0.953; Figure 6). The maximum values appeared in the soil depth of 0-5 cm, in which the cohesion of Kochia scoparia was 8.04 kPa, while that of Artemisia sacrorum was 5.18 kPa. There was no difference in the cohesion between the two plants in different soil layers below a 10 cm soil depth. The cohesion of the 0-10 cm soil layer was significantly different from that of the 10-30 cm soil layer, and that of the 0-10 cm soil layer was significantly higher than that of the 10-40 cm soil layer. The average root cohesion of the two plants was Kochia scoparia (2.73 kPa) > Artemisia sacrorum (1.60 kPa) plants in different soil layers below a 10 cm soil depth. The cohesion of the 0-10 c layer was significantly different from that of the 10-30 cm soil layer, and that of th cm soil layer was significantly higher than that of the 10-40 cm soil layer. The a root cohesion of the two plants was Kochia scoparia (2.73 kPa) > Artemisia sacrorum kPa)

Root Morphology and Soil Depth
Root indexes such as root number, root distribution depth, root surface are root volume are important factors affecting soil shear strength [33]. In this stud roots of the two plants were mainly distributed in the shallow soil layer less than There was no significant difference in the number of roots in the shallow soil la 0-10 cm, and the number of roots in the upper soil layer was more than that in the soil layer. The root number of Kochia scoparia increased first and then decreased w increase in soil depth, which was similar to the root distribution of Vetiveria ziza (L.) Nash [37]. However, Artemisia sacrorum showed a decreasing trend. The reason be that the taproot of Artemisia sacrorum was developed, and there were no obvio eral roots. It is consistent with the research by Burylo et al. [38] on six species, inc two tree species, Pinus thunbergii Parl. and Quercus pubescens Willd., two shrubby s Genista cinerea D. C. and Thymus serpyllum L., and two herbaceous species, Achna

Root Morphology and Soil Depth
Root indexes such as root number, root distribution depth, root surface area and root volume are important factors affecting soil shear strength [33]. In this study, the roots of the two plants were mainly distributed in the shallow soil layer less than 10 cm. There was no significant difference in the number of roots in the shallow soil layer of 0-10 cm, and the number of roots in the upper soil layer was more than that in the lower soil layer. The root number of Kochia scoparia increased first and then decreased with the increase in soil depth, which was similar to the root distribution of Vetiveria zizanioides (L.) Nash [37]. However, Artemisia sacrorum showed a decreasing trend. The reason might be that the taproot of Artemisia sacrorum was developed, and there were no obvious lateral roots. It is consistent with the research by Burylo et al. [38] on six species, including two tree species, Pinus thunbergii Parl. and Quercus pubescens Willd., two shrubby species, Genista cinerea D. C. and Thymus serpyllum L., and two herbaceous species, Achnatherum splendens (Trin.) Nevsk and Aphyllantes monspeliensis. The root number of Kochia scoparia and Artemisia sacrorum showed an exponential functional relationship with soil depth. The variation trend of the root diameter with soil depth was similar to that of root number. With the increase in soil depth, the average single root diameter decreased as a power function. The decrease in the root surface area can reflect the contact and connection between the root and the soil, and the root volume can reflect the growth status of the plant roots and the space occupied in the soil. The larger surface area and volume of the root result in larger contact area with the soil, a higher proportion of roots in the soil and a stronger root-soil bond. These two parameters decreased with the increase in soil depth. The variation pattern of the root surface area and root volume of Kochia scoparia and Artemisia sacrorum in the soil layer was the same as that of other herbaceous plants, such as Artemisia desertorum Spreng. Syst. Veg. [33], and was also similar to that of some shrub plants, such as Nerium oleander L. [8]. The root cross-sectional area, root surface area, root volume and root area ratio of Kochia scoparia and Artemisia sacrorum met the exponential function relationship with soil depth, and the determination coefficient of the fitting relationship was high (R 2 > 0.9). Other studies also showed that root density declined exponentially with soil depth [39], indicating that the relationship between these four morphological indexes of herbaceous roots and soil depth can be expressed by the exponential function. However, it should be noted that the distribution of root area ratio has significant spatial heterogeneity, and the factors such as plant category, growing climate and soil environment will affect the root area ratio of plants to different degrees. There may be an exponential functional relationship between root area ratio and soil depth [40] or a trend of increasing first and then decreasing [3]. Therefore, the root area ratio at different soil depths should be carefully applied to evaluate root cohesion.

Root Tensile Properties and Root Diameter
The tensile force of Kochia scoparia and Artemisia sacrorum showed a power function or an exponential function with the increase in root diameter ( Figure 5), which was similar to Sustainability 2022, 14, 8053 9 of 12 the results of the relationship between the root tensile force and the diameter of herbaceous plants such as Paspalum notatum Flugge [41], and also similar to that of shrub plants such as Crataegus microphylla C. koch and Mespilus germanica L. [42]. The tensile strength decreased with the increase in root diameter in the power function. This relationship also widely existed in other shrub plants such as Rosa canina (L.), Cotoneaster dammeri (C.K. Schneid) and Juniperus horizontalis (Moench) [27] and other trees such as Fagus longipetiolata Seem., Tamarix ramosissima Ledeb and Larix decidua Mill. [43].
The tensile force of Kochia scoparia was lower than that of Artemisia sacrorum, but the tensile strength was higher than that of Artemisia sacrorum, which was closely related to the root diameter. As the tensile force increases with the increase in the root diameter, and the maximum tensile strength decreases with the increase in root diameter. The average diameter of the root system of Kochia scoparia was smaller than that of Artemisia sacrorum in this study. Root tensile properties are also directly related with the content of chemical components in the roots. Generally, the content of cellulose and lignin in the roots is positively correlated with tensile strength, while the content of cellulose and tensile strength in the roots is significantly negatively correlated with diameter [44,45]. A higher content of cellulose and lignin in roots can result in greater root tensile strength. For example, the tensile force of Carpinus betulus L. (95.36 N) is greater than that of Acer velutinum Boiss. (64.15 N), and its tensile strength (43.31 MPa) is also greater than that of Acer velutinum (30.77 MPa) [42]. In fact, the differences of average root tensile strength between the two herbaceous plants in this study and that of some trees and shrubs are not obvious, even higher than that of some trees and shrubs. For example, the average tensile strength of Kochia scoparia (18.34 MPa) and Artemisia sacrorum (15.84 MPa) was higher than that of shrub species Inula viscosa L. (11.52 MPa) [18] and tree species Fraxinus excelsior (12.74 MPa) [42]. Therefore, the roots of herbaceous plants have a potential advantage in reinforcing shallow surface soil.

Root Cohesion and Soil Conservation
The root cohesion of Kochia scoparia and Artemisia sacrorum was mainly concentrated in the 0-10 cm soil layer. With the increase in soil depth, the root cohesion of the two plants decreased (Figure 6). This result was consistent with the research by Burylo, Hudek and Rey [38]. However, other studies showed that the root cohesion of Lycium andersonii A. Gray first increased and then decreased with the increase in soil depth [5], which may be due to the differences between plant species and soil conditions, resulting in different distribution laws of root tensile strength and root area ratio in the soil layer.
According to the Wu-Waldron model, the average root cohesion of Kochia scoparia and Artemisia sacrorum were 2.73 kPa and 1.60 kPa, respectively. However, the Wu-Waldron model assumes that the roots are tightly bound to the soil, and all roots break at the same time when the soil is shearing [9]. In fact, the root system does not reach the ultimate tensile strength at the same time when the soil is sheared. As a result, the Wu-Waldron model overestimated the soil reinforcement of the plant roots [46,47], which led to a larger calculation result of the root cohesion. Research has shown that the actual soil reinforcement effect of roots is only 34-50% of that calculated by the Wu-Waldron model [46,48,49], and its high value is positively correlated with root area ratio [50]. Therefore, in this study, the average root cohesion of Kochia scoparia may be 0.92 kPa-1.37 kPa, and that of Artemisia sacrorum may be 0.54 kPa-0.8 kPa.
The loess structure is loose and is easily scoured by water, which is liable to soil erosion and rainwater infiltration. Moreover, rainfall irrigation increases the soil water content of the slopes and leads to the deterioration of loess strength, and then induces loess landslide. The study proved that the root system can improve the shear strength of soil by exerting its own tensile characteristics. Chen [51] studied the difference in shear strength between the soil samples with roots and the unmodified loess through indoor direct shear and model tests. The results showed that the strain of the samples with roots reached the maximum shear strength, which was significantly higher than that of the samples without roots, and the existence of roots increased the internal friction angle and cohesion of the loess. Su et al. [33] found that the root system of Artemisia sacrorum had a significant influence on the safety factor (Fs) of the loess slope. With the increase in soil depth, the roots had a gradually decreased influence on the slope Fs. In the upper 0-0.3 m range of the soil layer, the effect of roots was obvious, and the presence of roots could significantly improve Fs. Although some trees and shrubs have stronger roots than herbs, which results in better soil reinforcement [18], they are difficult to grow on steep slopes due to the need for growth conditions; they consume more water and nutrition than herbs [52,53]. Therefore, herbaceous plants are a competitive selection for loess slope ecosystem reconstruction and slope stability protection.

Conclusions
Based on the root morphology investigation, root tensile tests and the root cohesion evaluation of the two typical herbaceous plants, Kochia scoparia and Artemisia sacrorum, the main conclusions are as follows: (1) The root morphological indexes of the two plants decreased with the root depth in the power functions (root diameter) or exponential functions (root number, root crosssectional area, root surface area, root volume and root area ratio). These morphological distribution indexes can comprehensively reflect the root distribution in different soil depth. It can be inferred that Kochia scoparia and Artemisia sacrorum are suitable for strengthening and enhancing the stability of the shallow soil of the slope. The soil reinforcement effect of Kochia scoparia is obviously higher than that of Artemisia sacrorum, which can be considered the dominant herb species in the ecological slope protection project.