The Pullout Mechanical Properties of Shrub Root Systems in a Typical Karst Area, Southwest China

Abstract

Roots play a major role in reinforcing and stabilizing soil. The pullout mechanical characteristics of soil reinforcement and slope protection of the root systems of dominant shrub species (Pyracantha and Geranium) were estimated by in-situ pullout tests in a karst area, in which roots were pulled out from soil to reliably test the pulling force. The goals of this study were to discover the pullout mechanical properties of roots in karst areas and to try to analyse the impact of the root system on landslide control. The F–s curves were multipeak curves with a noticeable main peak and main double peaks. The curves showed a linear increasing trend at the initial stage of drawing and decreased rapidly after reaching the peak. The F–s curves of root systems inserted into rock cracks showed secondary fluctuations in the later stage of drawing, and rock cracks stimulated the tensile efficiency of the root system more effectively. Field in situ pullout results indicate that tree roots fail progressively rather than simultaneously. The maximum pulling force had a linear relationship with the increase in soil thickness and a disproportionate increasing trend with the increasing number of broken roots. The displacement of the maximum peak was different between the two tree species and was concentrated at 5–15 cm and 5–25 cm for Pyracantha and Geranium, respectively. The maximum pulling force of Geranium was 1.29 times that of Pyracantha, and the root system of Geranium had strong pullout resistance. We concluded that the peak distribution of the F–s curves was affected by broken roots and rock cracks, while soil thickness and the number of broken roots had positive effects on the maximum pulling force, all of which is helpful in understanding the effect of root pullout mechanical properties on landslides in karst areas.

1. Introduction

As a representative area of the karst plateau in China, Guizhou Province is known as the karst province in China. Due to the interaction of natural environments (rainfall and geological geomorphology of carbonate rocks) with man-made disturbances and damage, this area suffers from severe soil erosion, large-area bare bedrock and degradation in land productivity [1]. Soil loss is the main factor causing karst rocky desertification, considered one of the most serious ecological, geological and environmental problems in Southwest China. Long-term karstification forms the surface and underground double layer space structure in the karst region, in which the surface is rugged complex terrain, and the underground is a large underground river system. Under the interaction of natural factors (rainfall and geological geomorphology) and human activities, the surface presents a rocky desertification landscape with a shallow soil layer, discontinuous regolith and even large areas of bare rock. It has been posited that soil loss on karst slopes includes ground and underground soil losses [2]. Therefore, landslides are more likely to occur in karst areas due to the complex heterogeneity of underground space (Figure 1). Landslides are a pervasive natural hazard in alpine regions [3] that threaten the built and natural environment [4]. Expressways have made great progress in Guizhou Province, but due to the massive excavation of slopes, human activities have caused great soil erosion and the occurrence of shallow landslides Shallow landslides constitute one of the most hazardous categories of mass movements, mainly because of their frequent evolution into rapid mass movements, characterized by debris avalanches and flows [5]. The great role of vegetation in stabilizing hillslopes is well established [6,7]. Vegetation strongly affects mechanical and hydrological soil behaviour, particularly related to shallow landslides. This is mainly through the following aspects: (i) influence on soil suction by root water uptake; (ii) reinforcement of the soil due to the presence of roots, increasing the tensile strength; (iii) anchoring of the shallowest layers to the deep and usually more stable substrates; (iv) surcharge due to weight of plant biomass (aerial part and root system) that increases the normal stresses to the slope; and (v) rainfall interception by canopy and evapotranspiration, reducing the delivery rates of intense precipitation and lowering the water table [8]. Root tensile strength and stiffness resist shear to provide mechanical reinforcement to the soil [9]. At the hillslope scale, the presence of vegetation generally increases soil thickness, lowering the frequency of landslide events [10]. Tree roots are regarded as a factor determining the stability of hillsides and riverbanks and can prevent the direct inflow of rainfall on the slope’s surface, thus providing stability against surface layer loss and scouring [11] and delaying the process of erosion and massive waste [12,13,14]. Plant roots can enhance slope stability [15,16,17], and their effects are exerted through basal root reinforcement and lateral root reinforcement [18]. Different constraints for the efficiency of soil fixation capacity exist between basal root reinforcement and lateral root reinforcement: basal root reinforcement plays a role when roots cross the shear plane, and the root reinforcement results are controlled by large roots [19]. The significance of mechanical slope stabilization by roots hence primarily depends on the thickness of the potential slip surfaces, the likely failure mode and the steepness of the slope [20]. Lateral roots reinforce soil on hillslopes and are limited in landslide areas because their contribution to slope stability is proportionally reduced by increasing landslide size [21]. Coarse roots anchor the soil through the friction force at the root–soil interface, and the anchorage effect of roots strongly depends on thickness and spatial density [22]. Fine roots enhance the shear resistance of soil. Roots reinforce soil, and the distribution of roots within vegetated slopes strongly influences the spatial distribution of soil strength and root pullout force [20,23]. Horizontal fibrous roots and vertical roots have varying degrees of effectiveness in fixing soil on the slope [24]. Vertical root systems strengthen soils’ resistance to shear by the way in which the roots cross the shear surface [25], whereas horizontal and shallow root systems have strong interaction effects with adjacent root systems [18]. In addition to the above factors affecting root soil fixation, the presence of rock cracks in the soil may have a great influence on the mechanical traits of roots. Roots grow downward and are anchored to soil, topsoil, and/or bedrock due to geotropism. Roots can penetrate bedrock joints and fractures, achieving strong and deep anchoring for the plant [26,27]. At present, most studies focus on the mechanical traits of single roots and investigate factors affecting the mechanical properties of single roots. However, the superposition of the mechanical behaviour of single roots fails to accurately characterize the slope protection effects of tree roots. Therefore, quantifying all root-pulling forces of a tree by an in-situ pull test is of great significance to slope treatment.

In recent decades, the assessment of landslide-prone areas has become one of the most discussed topics in the literature because of the difficulty in predicting landslide events due to their complex nature [28]. Geological complexity, geomorphological deformations and landuse and landscape changes are the main causes of the uncertainties around landslide occurrence [29,30]. Vegetation measure is an effective way to prevent soil erosion [31] and shallow landslides. Plant roots, as media of communication between plants and soils, exert mechanical effects that include the reinforcement and anchoring of soil, and improvement of the stability of slopes [32]. Thus, it is necessary to further explore the root pullout characteristics in landslide control. Slope stability models include the effects of roots by adding an apparent cohesion to the soil to simulate root strength [33]. To understand the mechanical properties of the root system, some researchers have introduced model applications. The Wu [34] model was used to calculate additional cohesion from the roots (Cr), which overestimated the actual soil reinforcement by roots because of the assumption that all roots break at the same time regardless of the diameter. The extended root bundle model (RBM) is different from the Wu model in that root breakage is considered as progressive breaking, which contains the roots’ critical parameters, such as force and tortuosity, concerning root diameter. The root system can effectively improve the shear strength of soil, the WW model was adopted to estimate the rise in the shear strength of soil due to roots [25,34,35]. These models were seen in

Table 1. The advantages and disadvantages of mentioned models.

Model	Model Name	Creator	Function	Advantage	Disadvantage
${\mathrm{c}}_{\mathrm{r}}={\displaystyle \sum }{\mathrm{F}}_{\mathrm{i}}{\mathrm{N}}_{\mathrm{i}}$	Wu	Wu	To calculate additional cohesion from the roots (Cr).	Simple to use	Overestimates the actual soil reinforcement by roots
$\mathrm{F}={\mathrm{A}}_{\mathrm{o}}{\mathrm{d}}_{\mathrm{x}}={\mathrm{ad}}^{\mathrm{b}}$	RBM	Schwarz	To estimate the maximum root resistance force	Root breakage is considered as progressive breaking and close to the true state of broken of roots	The model parameters still need to be tested in-situ or in the laboratory.
$\mathrm{s}=\mathrm{c}+{\mathrm{s}}_{\mathrm{r}}+\mathsf{\sigma }\tan \mathsf{\phi }$	WW	Wu and Waldron	To estimate the rise in the shear strength of soil due to roots	Simple and rapid	Study scale is limited

Where F_i is the root pullout resistance force of the diameter class, i and N_i is the number of roots in the diameter class. Where x is the root force shape factor, which can be calculated as b⁻², and A_O is the root force scaling factor, which is calculated by a × 4/π due to roots (kPa), σ represents normal stress on the shear plane and φ is the angle of internal friction (◦). Where S represents total shear strength of root–soil complex (kPa), c represents soil cohesion (kPa), S_r represents increased shear strength.

. Additional apparent cohesion from plant roots is of utmost importance in some physically based models, and lateral root strength is a primary control of the landslide size and location. Pullout tests are mostly used to estimate root reinforcement [36] and include single pullout tests in the laboratory and in in-situ pullout tests. Some achievements have been made in single root mechanics; for instance, it has been established that the maximum tensile force of a single root has a positive correlation with root diameter [37] and that root tensile strength decreases exponentially with increasing root diameter [27,38,39]. Although the mechanical traits of a single root can reflect the mechanical properties of the root system to a certain extent, they incompletely explain the soil reinforcement ability of the whole tree. The pulling force measured by an in-situ pullout test can not only reflect the interaction force between roots and soil more accurately but can also truly describe the mechanical properties of plant roots under natural conditions [40]. As an important factor in evaluating slope stability, the mechanical traits and soil fixation ability of root systems play an important role in reducing landslides and controlling soil erosion on slopes. This work presents significant research about the connection between landslide and root pullout mechanical characteristics. Many studies have discussed the mechanical behaviour of single roots and their influencing factors; nevertheless, exploration of uprooting characteristics in karst areas has seldom been carried out. This study fills this gap by evaluating and quantifying the pullout force and influencing factors for Pyracantha and Geranium. For the issues mentioned above, the mechanical properties of single roots and root bundles are different in terms of slope protection and soil consolidation. The specific goals of this study were to (1) analyse the characteristics of the relationship curves between root pullout force and displacement value (F–s curves) and affected factors on the peak distribution of F–s curves, (2) to determine the influence of soil thickness and root breakage on the maximum pulling force, and (3) comparatively analyse the pullout resistance of two tree species.

2. Materials and Methods

2.1. Materials

The study area was located in Huaxi District, Guiyang City, Guizhou Province, with a geographical location of 26°27′11″ N, 106°39′3″ E and an altitude of 1130 m, Huaxi District is a typical karst geological area. The topography of the karst mountainous area is undulating, and landform types such as peaks, mountains, hills, depressions, valleys and basins are diverse. The geology consists of huge thick marine carbonate formations. The lime soil is formed by the weathering of carbonate rocks in the karst area, and its clay minerals are mainly kaolinite and vermiculite, with slow soil formation rate, heavy texture, poor permeability, strong expansion and contraction, and poor corrosion resistance. The soil types in the area are diverse, including yellow soil, lime soil, red soil, brown soil, rocky soil, etc. The soil in the study area is shallow, with a soil layer thickness of approximately 6 cm, and the soil layer contains gravels of varying sizes. The bedrock exposure rate reaches 45%. Vegetation is a key factor in preventing soil erosion and promoting ecological restoration in karst mountainous areas. Revegetation and restoration have become fundamental measures to control the development of soil erosion and shallow landslides.

A landslide is defined as the mass movement of soil/earth or debris down a hillslope triggered by both natural (rainfall and geology/lithology) and anthropogenic (hill cutting and deforestation) causes [41]. Landslides occur when slopes undergo a decrease in the shear strength of the hillside material due to an increase in the shear stress, or due to a combination of natural ecosystem processes and anthropogenic activities [42]. According to the nature of rock and soil, landslides are divided into two categories: bedrock landslides and loose layer (soil) landslides. The latter category, comprising shallow landslides covered with vegetation and broken rocks in karst areas, formed naturally after heavy rainfall, was analysed in this study. Climatology, geomorphology, geology, geo-structure, seismic activity, landslide prone areas and man-made activities are the main triggering factors for landslides [43,44]. Landslide disasters are associated with anthropogenic factors such as environmental degradation, unsustainable development planning, cultural barriers and lack of community risk perception and good governance [45]. Landslides generally occur in mountainous areas due to their steep slopes. Topography and hydrology influence debris flow initiation through the effect of gradient on slope stability with rainfall [46]. Geological-based triggering factors play important roles in determining landslide occurrence in Isfahan [30]. Heavy rainfall is one of the main causes of shallow landslides [47], high soil moisture content reduces internal soil cohesion, which increases the likelihood of landslide damage occurring. The study area belongs to the subtropical humid monsoon climate zone and the average annual rainfall in this area is 1200 mm. Slope landslides are prone to occur in the rainy season, and erosion gullies are prone to form in areas without vegetation coverage.

2.2. Tree Species Selection and the Characteristics of Experimental Plants

Pyracantha and Geranium were selected as the experimental trees since they are typical revetment shrubs with a well-developed root system in the study area. According to a survey, they have both vertical and horizontal roots, which meet the needs of root slope protection. Before pulling the samples out of the soil, the ground diameter (height of the trunk 5 cm from the ground), plant height and crown area were measured. The thickness of the soil in which the experimental plants were growing was measured by a ruler, and the soil thickness at tree growth locations was recorded as 0 in places that were rocky or where the soil was too thin to measure. All the measured data are reported in

Table 2. Basic characteristics of experimental plants.

Species	Ground Diameter (mm)	Plant Height (cm)	Soil Moisture	Crown Area (m2)
Pyracantha	22.2	130	26.7	0.36
	25.00	65	24.6	0.80
	22.48	130	12.2	0.30
	23.12	190	26.3	0.72
	29.16	220	26.2	3.06
	25.04	130	26.2	0.66
	24.30	160	27.6	1.00
	17.60	130	25.3	1.10
Geranium	25.10	167	28.3	0.42
	24.52	210	29.6	0.49
	26.52	180	26.8	1.10
	18.22	168	15.3	0.72
	22.80	245	27.8	1.01
	17.88	200	41.3	0.56
	26.64	450	41.3	0.48
	19.20	210	42.3	0.90

.

2.3. Device and Experimental Procedure

The pulling force of the whole plant was measured by a whole-plant root puller (designed in the North Campus of Guizhou University). The experimental device consisted of a tripod, pulley block, wire rope, metal fixture, force sensor and displacement sensor, digital display and jack (Figure 2). The main portion of the experimental device was the tripod. After the device was assembled, the metal fixture was fixed at a height of 5 cm from the ground, and the range of the fixture was 18~40 mm. The load was applied by means of a jack, the drawing rate was 10 s/cm, and the load was transmitted through the pulley. Changes in drawing force and displacement were recorded by the digital display, which was connected to the two sensors. Pulling force data were recorded from when the displacement changed by 1 cm until the whole tree was pulled out. A tape measure was used to verify the change in displacement. The maximum load value of the digital display force gauge was 5 kN, the scale value was 0.001 kN, and the measurement accuracy was 0.5%.

3. Results

3.1. Characteristics of the Relationship Curves between the Root Pullout Force and Displacement Value (F–s curves)

The F–s curves showed different trends and were divided into two categories according to the characteristics of the peak distribution (Figure 3). The peak occurred when the root system was pulled out or broken. All curves had a consistent trend in which the pulling force increased linearly with increasing displacement at the initial drawing stage and the curves dropped sharply in an undulating pattern after the maximum peak appeared. The type a curve was a multipeak curve with a noticeable main peak that declined rapidly in a wave form after reaching the maximum (Figure 3a). The type b curve was similar to the type a curve; it was a multipeak curve with main double peaks, one of which moved back in the late stages of displacement. The second peak of the curve occurred when the root system penetrating into the rock cracks was pulled out (Figure 3b), something which is a common root–rock coupling phenomenon in karst areas.

From the F–s curves of the root systems produced from the pullout test, the maximum force during the pullout test and the peak displacement could be obtained. The peak displacement indicates the displacement at which the root will slip or break when the pullout force reaches the maximum. The peak displacement values of the maximum peaks differed due to the characteristics of the two tree species. The maximum pullout force and the peak slippage could be obtained from the F–s curves of the roots, as shown in Figure 3. The peak displacement of Geranium was concentrated at 5–15 cm, and that of Pyracantha was concentrated at 5–25 cm, which indicates that the root system of Pyracantha had greater deformability and effectively resisted the occurrence of shallow landslide displacement. The mechanical properties of root tension can effectively reduce the harm caused by shallow landslides.

3.2. Influence of Soil Thickness on Maximum Pulling Force

Soil thickness is particularly important for the mechanical properties of a root system. The soil thickness in the study area is unevenly distributed, and samples with soil thicknesses of 0–30 cm were selected for testing to study the pulling force of the whole tree root system. The maximum pulling force was related to the soil thickness. The deeper the soil thickness is, the larger the contact area between the roots and soil, and the contact area between the roots and soil has an important influence on the root reinforcement of soil. The maximum pulling force of Geranium first increased in the soil thickness range of 0–20 cm and then decreased with increasing soil thickness, and the optimal soil thickness ranged from 11 to 20 cm. For Pyracantha, the maximum pulling force tended to increase with increasing soil thickness. A soil thickness of 0 meant that the environment in which roots grew was covered by stone. The symbiotic environment of mixed stone with soil greatly reduced the mutual contact area between roots and soil. A soil thickness of 0 corresponded to the minimum value of the maximum pulling force for the two species of trees. The maximum pullout resistance of Geranium was stronger than that of Pyracantha at similar soil thickness values (Figure 4).

3.3. Resistance Response of Broken Roots to the Maximum Pulling Force

The effect of root reinforcement depends on the tensile strength of the roots and the bonding strength between the soil and roots. In in-situ pullout tests, the number of roots broken corresponds to the efficiency of the root system in resisting external stress in the process of slope failure. The variation trend of the maximum pullout resistance with the increase in the number of broken roots is shown in Figure 5. The maximum number of broken roots for Geranium is 29, and for Pyracantha is 18. With the increase in the number of broken roots, the maximum pulling resistance had a disproportionate increasing trend. The number of root system breakages of Geranium was greater than that of Pyracantha.

3.4. Different Characteristics of Root Drawing between the Two Tree Species

According to systematic measurements and statistics, the number of broken roots was higher in Geranium (mean = 20.5 ± 4.50) than in Pyracantha (mean = 10 ± 5.81). The average pullout force and average pullout displacement refer to the average of all data recorded during the process of a tree being pulled out. As shown in Figure 6 the average pulling displacements of Geranium (mean = 58.13 ± 15.59) were larger than those of Pyracantha (mean = 43 ± 18.97). The average maximum pulling force of Geranium was 1.29 times that of Pyracantha.

The distributions of the maximum and average pulling forces of the two species are presented in Figure 7. Root reinforcement depends on the ability of roots to resist pulling. The maximum pulling force of Geranium was generally greater than that of Pyracantha, and the average pulling force and the maximum pulling resistance had the same distribution between the two species. The maximum pullout resistance of the whole root system of Pyracantha ranged from 1.12 to 3.12 kN, and the maximum pullout resistance of Geranium ranged from 1.58 to 4.32 kN. Regarding the root resistance of the whole tree, Geranium had better pullout resistance.

4. Discussion

4.1. Characteristics of F–s curves

The F–s curves of the two tree species were divided into two categories according to the characteristics of the peaks, and they were multipeak curves with a noticeable main peak and main double peaks. The curves had a similar trend in the early stage of pulling: the pulling force increased linearly with displacement and then dropped sharply in an undulating pattern after the maximum peak appeared (Figure 4). The difference between the two kinds of curves lay in the late drawing period: the type b curve showed obvious quadratic fluctuation in the later period due to the root system penetrating into rock cracks. The presence of rocks in the soil would change the nature of the root–soil interface, which could reduce the contact area between roots and soil, greatly reducing the friction at the root–soil interface. However, when roots are subjected to pulling force, their tensile strength is more effectively stimulated if the roots penetrate into rock cracks. In this study, a second obvious peak appeared at the end of the type b curve, and the root system exhibited greater pullout resistance. The existence of rock cracks can stimulate the tensile properties of roots to a greater extent and strengthen the ability of roots to solidify soil. The efficiency of root reinforcement depends on the pullout resistance of the root system. The pullout resistance was mainly affected by the tensile strength of the root system, the interaction area and the roughness of the root–soil interface. The root system was no longer easily pulled out and deformed more when the root system penetrated into rock cracks, which stimulated greater pullout resistance in root reinforcement. Soil reinforcement by roots depends on the root tensile strength. When the force on a root exceeds its tensile strength, the root will be broken or pulled out. When the root system was broken or pulled out, a peak appeared in the F–s curve, and the contribution of the root coefficient to the drawing force exhibited a nonproportional increase [48], because single root sliding or dislocation and loosening of soil particles would cause the load to be redistributed. The F–s curves were similar to the P–s curves from the study of Chen Lihua [40] and were unimodal and bimodal. The two shrub species selected have both vertical and horizontal root systems, and root systems pulled out by force would not be broken at the same time. The activation of root strength within bundles was not synchronous, and progressive root failure must be considered when quantifying effective root reinforcement [49].

4.2. Influence Mechanism of Root Failure Modes on Root Mechanical Properties during Pullout

Two failure modes may exist in a root pullout test: pullout roots and breaking roots [36]. In this study, the two root failure forms are presented, and schematic diagrams of different root failure modes are displayed for Geranium and Pyracantha according to field surveys in Figure 8. The number of root system breakages show that Geranium roots were more likely to be broken than that of Pyracantha. Diverse plants have different contents of internal chemical components, cellulose, lignin, hemicellulose, etc. in their roots, which causes a large difference in the mechanical properties of the roots [50,51]. Field observations showed that in tree-root bundles, the dominant failure mechanism of roots was breakage [52]. Slippage was limited to small roots that usually contributed only a small fraction of the total root reinforcement. Because of the high heterogeneity of the soil environment in karst areas, there is no specific statistical law relating slip out and root fracture. In addition to the influence of the soil environment on the stress–strain behaviour of the root system, the characteristics of the root system play an important role; such characteristics include root tortuosity [20], root–soil mechanical interactions [52], the position of root breakage along the root, branching, root geometry, changes in root diameter, and the presence of stones. The presence and position of branch roots strongly affect the peak pullout resistance. Deeper branch roots can result in greater pullout resistance [53]. However, due to the limitation of the in-situ pullout test, root branches have not been considered. Root pullout resistance was mobilized gradually, and roots failed at different amounts of displacement, depending on their individual morphology [54]. Roots were pulled out or broken depending on the interaction between the root and soil and the root tensile force. When roots are pulled, if the force required to break the root–soil friction bond is less than the force required to break the root, it will be pulled out of the soil, otherwise the root will be broken [55]. The friction between the soil and root determines the type of mechanical failure: in cohesive soils, small roots tend to break under dry conditions and slip out under wetter conditions [46]. Root pullout force is affected by the soil in which roots are embedded, such as soil type and soil water content [27,55] Testing hydrated roots will give a lower estimate of strength than testing dried roots, but it might overestimate root ductility and underestimate the time of slope failure [56,57].

4.3. Applicability of the Pullout Test

There are three common pull tests–pullout test of the whole plant, also named uprooting test [58,59], pullout test on remoulded soil-root samples [51,60,61] and in-situ pullout test [15,55]—that are often used to study root pullout mechanical properties. This study has some limitations that could be considered in future research. For instance, we used an in-situ pullout test to obtain the pulling force of the whole root system of the two tree species. The uprooting test can evaluate the pullout force of the plant, but it is not able to obtain the pullout force and pullout strength of single root. The remodelling specimens used carefully selected root segments, and the degree of bonding between the root segments and the soil was very different from the natural conditions. In field uprooting experiments, it is difficult to control the root age and physical parameters of the soil. The in-situ pullout test was carried out on undisturbed soil. Before the test, the bonding and interweaving state between the original soil root and the soil was the natural state. The method adopted in this research should reflect the interaction between a single root and the surrounding soil more realistically than the remoulded soil pullout test. The in-situ pullout test is the most important test to estimate root reinforcement by the root bundle model [62].

5. Conclusions

The pullout mechanical properties and influencing factors of the root systems of two shrub species were studied by an in-situ whole-plant pullout experiment. The results show that the F–s curves were multipeak curves with a noticeable main peak and main double peaks. The curve of drawing force showed a linear increasing trend at the initial stage of drawing and a rapid decrease after reaching the peak until all the roots were pulled out. The secondary fluctuation at the later stage of the bimodal curve was caused by the pulling out or breaking of the root system, where part of the root system penetrated into rock cracks. The existence of rock cracks can stimulate the tensile properties of roots to a greater extent and strengthen the ability of roots to solidify soil.

The pulling force was positively correlated with the increase in soil thickness. The maximum drawing force had a disproportionate increasing trend with the increase in the number of broken roots, and the number of broken roots was higher in Geranium than in Pyracantha. The average pulling force and the maximum pulling resistance had the same distribution between the two species. The maximum pulling force of Geranium was 1.29 times than that of Pyracantha. The root system of Geranium had strong pullout resistance. This study provides a reference for the selection of soil and water conservation tree species.