Direct Shear Strength on the São Francisco River Bank, Northeastern Brazil, With or Without Roots of Different Native Species

Several plant species have been studied as reinforcement elements against landslides at slopes, either to protect against the splash effect offered by shoots or anchoring the soil by the root system. The objective of this work was to investigate the influence of roots from shrub species over the soil mechanical attributes such as cohesion and angle of internal friction at the São Francisco riverbank, northeastern Brazil. A trench was excavated under the treetop of every shrubby individual, sampling blocks at 0-0.25, 0.25-0.50, 0.50-0.75, and 0.75-1.00 m depths. The moisture contents, particle size, liquidity limits, plasticity and actual specific mass of the samples were determined. The soil physical and mechanical attributes such as initial and final specific weight, initial and final void content, initial and final degree of saturation, shear resistance peaks, cohesion, and internal friction angle were identified through direct, elementary, consolidated, and undrained tests. In the samples with Solanum paniculatum, Mimosa pigra species, the highest values for cohesion were registered for the samples with roots. Sesbania virgata presented the greatest variation in cohesion and internal friction angle among samples with and without roots. The position and root status influenced the results of the direct shear tests.


Introduction
Riverbank slopes can be stabilized by controlling the erosion process requiring the use of techniques that can be widely adapted to water basins in order to promote aesthetic and ecological improvements and to guarantee agricultural production in non-degraded areas associated of riverbanks (Abedini, Said, & Ahmad, 2012).
In some points of the sedimentary stretch on the lower Sã o Francisco Riverbank, the erosion magnitude rate can vary between 5 to 300m per year (Holanda et al., 2007;Rocha et al., 2018). The margin retreat lines are directly related to the energy of the natural waves, caused by the winds, soil mass movement on the slopes and hydrological events, as well as the variation of the flow, the variation of the height and the velocity of the flow (Ribeiro et al., 2011;L. Wang, Guo, & Wang, 2020). This, associated with the physical attributes of the slope in each section of the river, leads to the larger erosion rates at certain times of the year and in specific stretches (Martins-Oliveira et al., 2020). As mentioned by Holanda et al. (2005) it causes several consequences changing environmental landscape and reducing the crop area of irrigated agro ecosystems, dominated by small farmers and affecting the river navigation.
The use of slope recovery plants, mainly for geotechnical stabilization, demands characteristics that allow their development in these places. These characteristics include rapid growth, drought tolerance, deep root system, vigorous growth, wide availability of seeds, easy dissemination, survival in soils with low fertility and effective soil cover (Hamidifar et al., 2018;Lyda et al., 2017). Some species have been widely studied as important alternatives to control the erosion process on slopes and river banks, because their root systems can provide stability (Vannoppen et al., 2017).
The root systems contributes to the stability of the slopes, works in special emphasis on the mechanical effects (Kokutse et al., 2016;Nguyen et al., 2019). When soils are subjected to shear effects, roots mobilize their tension resistance and act as anchors, then the shear stresses in the soil matrix are transferred to the roots , which reinforces the soil mechanical resistance to tension (Correa et al., 2019;Leung et al., 2018).
Soil shear resistance (r), expressed by Coulomb's equation, is a function of the cohesion and the friction angle among the soil parameters (Lambe, 1951), which are intrinsic to the soil and are defined by their properties and attributes such as texture, structure, organic matter content, specific weight, mineralogy, and moisture content. Several authors agree that a plant root system may significantly influence the parameters related to the soil shear resistance (Ghestem et al., 2014;Stokes et al., 2014). Velinova et al. (2019), stated that plant roots tend to gather with the soil to form a monolithic mass that contributes to an increase in resistance, which generates an apparent cohesion. Thus, if the soil contains roots, the increased resistance to the soil shear may be expressed as additional cohesion. Mao et al. (2012), reported that the density and architectural characteristics of roots are likely more important than their mechanical cohesion characteristics. However, many authors support the theory that shear resistance in the soil is primarily a product of stiffness and tensile strength of the roots (Alam et al., & Baral, 2018;Ganapathy et al., 2019;Moshi et al., 2020). The resistance of soils that contain roots may be inspected directly in situ (Baranska et al., 2005;Carvalho| et al., 2020) or through laboratory tests such as direct shear tests (Kokutse et al., 2016;Qi & Tang, 2018).
Despite of large plant species biodiversity at the Lower Sã o Francisco River, there is a lack of information about which ones can be used in soil bioengineering designs, considering the suitable species features related to its mechanical properties. Thus, the objective of this work was to investigate the influence of roots from native shrub species over the soil mechanical attributes such as cohesion and angle of internal friction at the Sã o Francisco riverbank, Brazil.

Study Site
This study was conducted at the experimental site located in the banks of the Sã o Francisco river, in the municipality of Amparo de Sã o Francisco, Sergipe state, Brazil ( Figure 1). This bank is composed of sedimentary alluvial soil classified as Fluvic Entisol by the Soil Taxonomy (Soil Survey Staff, 2014). This region has a warm and humid climate with an annual mean temperature of 25°C. Rainfall ranges from 800 to 1300 mm in rainy season occurs during the winter, between March and September.
The research area was originally dominated by native riparian vegetation included in the Atlantic Rain Forest biome. During the last decades the original vegetation was gradually replaced by pasture and tillage.
In July 2010, soil-bioengineering techniques were installed in an experimental design. In these tests, intense soil mechanical operations mostly related to soil cutting and landfill in order to provide the slope reshaping resulted in extensive soil mobilization in the riverbank slope and in the associated flat area. This is about an area of 200 m × 30 m in length and width, respectively, and a slope of 27 degrees. During this period, the present vegetation, mostly composed by shrub invasive species, was clear cutted as a result of the slope reshaping, and the bank was planted partially with species such as vetiver grass (Chrysopogon zizanioides (L.) Nash) and other previously studied species to provide erosion control by its mechanical reinforcement. In 2012, prior to the beginning of this study, numerous plant species bloomed throughout the area due to the seed dissemination from the surrounding areas.

Species Selection
Some species were previously selected from the local biogeography (Franco, 1983), a plant survey conducted exclusively for shrub-sized plants. This botanical material was collected and preserved, and the species were identified in the ASE herbarium at Universidade Federal SERGIPE ALAGOAS Journal of Agricultural Studies ISSN 2166-0379 2021 de Sergipe (Nascimento, 2015). The survey resulted in a list of 10 potential species with desirable biotechnical characteristics. However, due to the availability of samples in satisfactory overall condition (Petrone & Preti, 2013) and time for excavation to collect soil blocks (Böhm, 2012), only the three most frequent species were selected: Solanum paniculatum, Sesbania virgata and Mimosa pigra. Solanum paniculatum L., a native Brazilian species commonly known as "Jurubeba", is an angiosperm of the Solanaceae botanical family. It is a perennial shrubby, erect, branched species 1 m-2 m tall exhibiting pubescent and curved spine stems, and its flowers have white petals and yellow anthers. Originally, from northern and northeastern regions, this species is present throughout the entire Brazilian territory, commonly found in pastures, crops, orchards, roadsides, tracks, and wastelands. It generally occurs in sandy and dry soils (Ferreira, Valente, & Santos, 2017). In soil excavations, this species roots achieve 0.63 m depth.
Sesbania virgata (Cav.), a native tropical species from Brazil, is part of the Fabaceae botanical family and has the botanical synonym of Sesbania marginata Benthan. Commonly known as "Sesbânia", it is a perennial, fast-growth shrubby plant. It is 2 m to 4 m tall and naturally occurs on riverbanks, floodplains, and modified soil (Sartori, Pott, Pott, & de Carvalho, 2018). This species produces a large quantity of long-term viable seeds dispersed in indehiscent fruits in the form of pods and has been used to re-vegetate riparian woods, soil erosion control, and recover degraded areas (Salis et al., 2004). In soil excavations, this species roots achieve 0.78 m depth.
Mimosa pigra L. is a native species of tropical environments and belongs to the Fabaceae botanical family. It is shrubby and erect and can reach up to 6 m in height. Its steam has thorns up to 7 mm long, its leaves are bipinnated with up to 16 pairs of 5-cm long pinnae. The seeds are able to germinate throughout the year and germination normally occurs in humid soil that has not flooded. It has a fast growth, and its flowering occurs 4 to 12 months after germination (Lonsdale, Harley, & Gillett, 1988). In soil excavations, this species roots achieve 0.87 m depth.
After selecting the individual with the best overall condition among each species, its root system was excavated and sampled and then submitted to direct shear strength tests.

Soil Sample
A trench was excavated under the treetop of every shrubby specimen, sampling four soil blocks measuring 25 × 25 × 25 cm, at 0-0.25, 0.25-0.50, 0.50-0.75, and 0.75-1.00 m depths from the soil surface, with a total of 12 blocks (four for each species). These blocks were paraffined and packed in wooden boxes to protect them during transportation.

Soil Characterization
The preparation of each intact block containing soil and soil permeated by roots and determination of the moisture content of the samples were conducted in accordance with the Brazilian Standard NBR 6457 (ABNT, 1986). The sieve analysis was conducted in accordance with the NBR 7181 Standard (ABNT, 1984d), and tests for determining liquidity and plasticity limits were conducted in accordance with NBR 6459 (ABNT, 1984a) and NBR 7180 (ABNT, 1984c) standards, respectively. The prescribed method in NBR 6508 (ABNT, 1984b) standard was used to determine the actual specific mass density.

Direct Shear Test
The soil physical and mechanical attributes such as initial and final specific weight, initial and final void content, initial and final degree of saturation, shear resistance peaks, cohesion, and internal friction angle were identified through direct, elementary, consolidated, and undrained tests (CU).
The tests were performed at the Geotechnical and Paving Laboratory (GEOPAV) of the Civil Engineering Department at 'Universidade Federal de Sergipe' in accordance with the D6528 (American Society for Testing and Materials, 2017) standard by employing a direct shear conventional press (Wille Geotechnik LO 2900). The horizontal displacement speed used in the tests was in accordance with the recommendation presented in the standard, which demands that the elapsed time necessary for obtaining the desired shear is twice that needed to obtain 90% consolidation of the species, which must be determined for each test. In this case, the established speed was superior to 0.05 mm min -1 . Therefore, in order to optimize testing times, this value was adopted as the standard shear speed in every test.
Four samples (blocks) containing only local soil and four samples (blocks) containing roots of the selected species were extracted of each intact soil block by using metallic rings (Ø = 60 mm × H = 20 mm), using a small blade to gradually to remove the soil excess. To simulate the average levels of effective geostatic vertical stresses in the studied slope, these species were submitted to normal stresses (σ) of 5 kPa, 11 kPa, 21 kPa, and 42 kPa, which were determined on the basis of the specific weight of the soil and the sample depth (Stokes & Mattheck, 1996).
The samples were saturated by flooding for 30 seconds. During the consolidation phase of each test, readings were made after 0. 1, 0.25, 0.5, 1, 2, 4, 8, 15, 30, 60, 90, and 120 min. During the shear stage, readings were made every 0.5 min until reaching 5 min, every minute until reaching 10 min, every 10 min until reaching 30 min and so on until reaching 30 min every 30 min. In this slow direct shear test with pre-consolidation, soil deformations were also measured ( Figure 2).

Figure 2. Sequence of procedures to perform the direct shear tests
After completing the tests, the resistance envelope was obtained for every applied normal stress by using Coulomb's equation. It was then possible to determine the values for the cohesion and internal friction angle of the soil for samples with and without roots.

Direct Shear Test
To evaluate the physical and mechanical attributes of the soil such as specific weight, cohesion, and internal friction angle, an experiment with a randomized block design (DBC) was performed that included two treatments (with or without roots), four blocks (depths: 0-0.25; 0.25-0.50; 0.50-0.75 and 0.75-1.00 m), and four block replications. The results were subjected to an analysis of variance and means comparison using Tukey's test with a 5% probability. Additionally, a correlation analysis between the resulting lines from the direct shear test in the samples with and without roots was conducted in accordance with the method described by Snedecor and Cochran (1989).

Soil Sample Characterization
The typical behavior of alluvial soils, which were formed by deposition of highly varied grain

Journal of Agricultural Studies
ISSN 2166-0379 2021, Vol. 9, No. 1 size layers (Gonzales, Dahlin, Barmen, & Rosberg, 2016;Jordanova, Goddu, Kotsev, & Jordanova, 2013), was not observed in the studied profiles. Although small variation between the soil constituents were reported, all of the blocks were predominantly formed by silt and were classified as CL under the Unified Soil Classification System, indicating inorganic clay with low compressibility (Table 1).

Direct Shear Tests
As expected, the sheaths of the root-permeated specimens exhibited different behaviors relative to those that did not contain roots, revealing an irregular rupture plane in those that contained roots and rectilinear in those that did not have roots (Figure 3). The species presented distinct behaviors in the effect of roots on the cohesion, internal friction angle, and initial specific weight for the different depths (Table 2). without roots. The values for c for samples without roots presented a little variation until the 0.75 m depth, at which the lowest cohesion of this study was observed at 0.07 kPa. Values close to zero are attributed to soils with extremely low clay content (Hubble, Airey, Sealey, De Carli, & Clarke, 2013), which was not the case for these samples (Table 2).
Conversely, ϕ was always lower for samples with roots except for the 0.25-0.50m depth, in which ϕ was lower for samples without roots. Samples with Sesbania virgata presented the greatest variation in c and ϕ among samples with and without roots. It is important to note that the lowest value for c, 0.5 kPa, was recorded for samples with roots at depths of 0.25-0.50 m.  (Table 1).

Samples with
While the tests were conducted, every sample, with or without roots, was submitted to the same moisture conditions of saturation. In addition, the contained soil in both sample groups (with or without roots) came from the same block; therefore, the unusual behavior can be caused only by the roots.
After the testing, it was observed that the sample with S. virgata, submitted to normal stress of 11 kPa, had a root segment 4 mm in diameter in the collapsed surface in the same direction as the shear. The sample of M. pigra, also submitted to normal stress of 11 kPa, showed decomposing thin roots.
The unexpected behaviors of the samples of S. virgata, at depths of 0.25-0.50 m, and M. pigra, at 0.75-1.0 m, both subjected to a normal stress of 11 kPa, may be verified by comparing the lower horizontal deformations with those samples subjected to normal tension of 5 kPa, for which lower horizontal deformation values were expected (Figure 4).

Journal of Agricultural Studies
ISSN 2166-0379 2021, Vol. 9, No. 1 For ϕ, however, the samples without roots had higher mean values, about 30%, than those of samples with roots (Table 3). Evidence for such effects may be observed in the linear regression line formed by the envelopes of samples with roots of S. virgata at depths of 0.25-0.50 m and M. pigra at 0.75-1.0 m when subjected to normal stress of 11 kPa.
Although linear regression lines presented good coefficients of determination at R² = 0.96 and R² = 0.95, the low peak resistance values attained by the samples underestimated the values of cohesion intercepts, consequently producing a higher value of ϕ.
Another significant aspect concerns the soil specific weight. Although widely constituted by silt and clay particles, samples that were extracted from blocks at greater depths presented lower specific weight values (Table 2), showing a significant difference between depths of 0.0-0.75 m and 0.75-1.0 m. Moreover, no statistical differences were observed between samples with and without roots.
The absence of a pattern explaining the effects of roots in the evaluated species may be verified in the envelope of shear resistance for each soil block at the studied depths (Figures 3 to 5).
For the Solanum paniculatum species, at 0-0.25 m depth, the lowest value and the shear resistance peak of the samples without roots were 9.75 and 31.85 kPa, respectively. The difference between these values justifies the slope of the regression straight line, which made it possible to determine the values of c and ϕ as 8.54 kPa and 30.81°, respectively for these samples. For samples with roots, the lowest value and the shear resistance peak were 9.1 and 18.85 kPa, respectively, for normal stresses of 5 and 42 kPa, which made it possible to determine the values of 8.1 kPa and 15.62° for c and ϕ. These results demonstrate that at such depths, both c and ϕ are lower when roots of this species are present ( Figure 5). A comparison of shear envelopes through correlation analysis showed no significant differences (P<0,50) in the lines from the samples with and without roots.
Evaluating the samples with roots, Sesbania virgata presented the highest shear resistance, 41.61 kPa, when submitted to a load of 42 kPa at depths of 0-0.25 m. Out of those without roots, Mimosa pigra presented the highest shear resistance, 46 kPa, when submitted to a load of 42 kPa at depths of 0.75-1.00 m.

Discussion
We believe that the predominance of silty soil is due to the mobilization of soil that occurred during the slope redesign in 2010, which altered its characteristics from natural deposition to embankment. Nonetheless, because it is an alluvial soil mainly in a sedimentary stretch, this silty layer could have been part of a soil particle grouping that was formed during a long deposition period with low water flow speed. Slaa et al. (2013), reported that silty soils are complex because silt is known as a cohesive material in erosion studies, but from a mineralogical perspective, it is classified as non-cohesive since it is composed primarily of quartz and feldspar.
2018; Pirnazarov & Sellgren, 2015). Nevertheless, it was observed that this relationship is fragile in samples randomly extracted from the site.
The premise that roots grow perpendicularly to the shear surface (Burak, Dodd, & Quinton, 2021;Meijer, Muir Wood, Knappett, Bengough, & Liang, 2019) could not be verified. In fact, the presence of roots allows a mobile slice of the sample to easily slide along the root, permitting the shear force to act with no resistance from the soil-soil friction; similar findings were presented by Campagnolo et al. (2018); Pollen and Simon (2010).
In the comparison of the average values of c between samples with and without roots, no statistical difference was observed for this attribute, which is the opposite to the values reported by Mao et al. (2012) and Palmer et al. (2016). These values for silty soil agree with those reported by  and are lower than those reported by .
During the sampling time, the studied species were three years old, it is possible that their root structures were not sufficiently developed for providing greater shear resistance in the soil. This would help to explain the lack of statistical differences in cohesion for soil samples with and without roots. et al. (2008), while evaluating the increase of shear resistance in situ, it was observed that the cohesion values for soils with roots exceeded those for soils without roots after the plants had fully developed.
The attained results show that ϕ was affected by the presence of roots, different than the findings of Mahannopkul and Jotisankasa (2019); Bischetti (2009).
Considering that the shear envelope is composed by angular and linear coefficients of Coulomb's equation (Lambe, 1951), any constituent that modifies the composition of the sample will also modify the value of these coefficients and, consequently, the values of c and ϕ.
Additionally, the presence of roots was expected to modify the behavior of the shear resistance envelope because its material was different from that of the soil sample. This corroborates the findings of Davoudi (2004), who observed a reduction in the values of ϕ in the presence of roots, which verifies a strong link between this attribute and the root diameter.
While examining the shear resistance of natural residual soil, which was a typical cohesive yellow lateritic soil, Silva and Carvalho (2007) observed a strong positive linear correlation between c and the soil specific weight, indicating dependency between these parameters for this soil type.
Nonetheless, the results of the present study show that soil specific weight is not linearly correlated to its cohesion or its internal friction angle, which is contrary to the results presented by Rocha et al. (2002).
This low structuring may be associated with the high concentration of silt, which is common in soils with less weathering (Ng et al., 2020). Moreover, the soil-soil friction in the shear surface, which increases the soil resistance (Amiri, Emami, Mosaddeghi, & Astaraei, 2019), was reduced due to the high voids content for samples at depths of 0.75 m to 1.0 m.
Biological factors such as decomposition of dead roots and the presence of tunnels, nests, and ISSN 2166-0379 2021 chambers to contain soil macrofauna such as insects, termites, and worms were observed at the laboratory during the sample preparation. Such factors are responsible for the decrease in soil-soil resistance.

Journal of Agricultural Studies
The position of the roots in relation to the shear plane influenced this behavior because, as previously stated, the samples were obtained from random positions within the blocks of soil. Therefore, the presence of roots was guaranteed although their direction, diameter, and level were not. Zhang et al. (2013) examined silty samples with roots of Robinia pseucdoacacia in three different positions to the cutting plane-perpendicular, parallel, and crossed-and concluded that crossed roots, followed by perpendicular roots, are most suitable for the soil reinforcement.
In this sense, because roots are compound less dense elements than the soil, it is expected that samples with roots have lower specific weights than those constituted only by soil, which influences their shear resistances (W. W. Rocha et al., 2002).
However, we should note that the method used by Comino et al. (2010) and Mao et al. (2013) for in situ determination of soil shear resistance is subject to environmental conditions. Therefore, this method considers the root system as a whole and not only as segments, as presented in this study. Thus, the apparent differences in this study can be attributed to the used methodology. The in situ method enables a higher contact between the soil and root surface. As a result, the friction and soil resistance are increased (Mckenzie, Mullins, Tisdall, & Bengough, 2013). Studies using this method to investigate shear resistance tend to present higher values of soil cohesion and resistance in samples with roots.

Conclusions
The different species provided varied behaviors in the effects of the roots on Cohesion values. Solanum paniculatum produced the highest values of c for samples with roots except for 0 m to 0.25 m.
As the soil appears without roots, the values of φ (friction angles) rise in comparison to soils with roots, approximately 30%, demonstrating that the presence of a cohesive root system influences the physical soil attributes. The species with the greatest significance was Sesbania virgata, which produced the most remarkable values of c for samples with and without roots. Mimosa pigra produced the highest c values for depths from 0.25 m to 0.50 m and 0.50 m to 0.75 m.
The soil specific weight is not linearly correlated to its cohesion or its internal friction angle.
It has been noticed the complex relationship between the presented data, mainly the expected relation between the root structure, and its contribution in the soil cohesion to the erosion control on the river bank.