3.1. Tests of Data Normality
Kolmogorov-Smirnov and/or Shapiro-Wilk attested the presence of normality, linearity (Table 2
) of clay, silt and sand across elevation classes. The null hypothesis is that if there was no significant difference among groups variance, it will be rejected by the Test of Equality of Error Variances. The p
-value for clay, silt, and sand across most of the elevation classes was 0.2. The test result assured the quality of the soil data set to proceed into ANOVA and linear regression analysis. Furthermore, Person correlation was run between soil texture (sand %) and bulk density (t/m3
) to validate the output of the laboratory analysis. The dependence test was significant (p
< 0.001) and the Pearson correlation value was 0.78.
The result of correlation analysis made between soil texture and bulk density attested the quality of soil sample data set. The finding was supported by the study of [15
]. That is, sandy soil has less number of pore spaces, while clayey soils have more pore space due to the huge number of tiny pore spaces.
3.2. The Association between Soil Texture and Terrain
The statistical output showed an association between spatial variability of soil texture (clay and silt) among slope and agro-ecological zones (elevation) groups (Table 3
). The extent of association was statistically significant (p
< 0.05). Unlike what was observed in the case of elevation classes, the distribution of silt was not correlated to slope classes. In normal circumstances, unlike clay and sand particles, the size of silt particles is not expected to correlate with terrain attributes.
Like the one-way-ANOVA, the regression output indicated the presence of a significant association (p
< 0.05) between soil texture (clay, silt and sand) and both terrain attributes. The standardized coefficient values of clay, silt, and sand attests that variation in texture was more pronounced among elevation values (Table 4
, Table 5
and Table 6
). In the estimation of clay particles, the standardized coefficients (Beta) value of slope and elevation were −0.499 and −0.767, respectively. Similarly, in the case of sand, the value of Beta for slope and elevation were 0.485 and 0.812 respectively.
The result of ANOVA and linear regression indicated the influence of terrain attributes over variation of physical soil properties among farmland plots. Referring to the USDA’s [17
] soil texture triangle, heavy clay, clay and clay loam are identified as the major texture classes in the lower, middle and upper parts of the watershed respectively. As indicated by Lark et al. [32
], the terrain attributes were significantly linearly related to physical properties of the soil. Similarly, in the finding of [23
], the soil texture varied from sandy loam to clay in the surface horizons down the slope. This could be associated with the removal of fine soil particles from steeper slope and their deposition at lower slope positions. The loose aggregate nature of the upslope soils has enhanced easy movement of the fine soil particles downslope along with the rain water movement [23
]. The finding of this study indicated that the catenary soil development in the watershed was well established. Most of the site is covered with clayey soil has poor circulation of water, nutrient and air in it. Tromp-van Meerveld et al. [16
] and [17
] were also stated that soil texture varied spatially with terrain attributes. As indicated by Cresswell et al. [30
], such soil texture class is characterized by low available water holding capacity for reasons associated with compaction.
The result of linear regression of this study showed the presence of higher influence of the agro-climatic classes on the spatial distribution of soil texture than that of the slope classes. The effect of slope value variation on texture was more pronounced among the agro-ecological zones of the watershed than within an agro-ecological zone. The result of both statistical analyses implied that the most recent and dominant geomorphic process in the watershed was fluvial. The result implies the presence of more noticeable effect of agro-climatic zone variation (i.e., an interactive effect of major factors like microclimate and biota) on the spatial distribution of soil texture than that of the effect of slope class variation.
Moore et al. [33
] explained the presence of differences in transport sediment particle size, whereby transport of course size particles was lower than that of fine particles. The studies made by [19
] explained that clay content increases with decreasing slope and downstream flow length and with an increasing upslope flow length. This study had discovered a positive correlation between elevation and sand particles.
Inconsistent to the finding of this study, areas with higher elevation presented higher values of clay content [22
]. On the other hand, Dessalegn et al. [23
] showed that the loose aggregate nature of the upslope soils has enhanced easy movement of the fine soil particles downslope along with the rain water movement. As reported by Kokulan et al. [34
], elevation, relative slope position and texture at the surface were highly correlated. That is, increasing sand content and decreasing clay content with elevation was observed. Sand content was higher in the upslope positions and lower relative to silt and clay in the downslope areas, Tesfahunegn et al. [35
] also presented that the highest values of fine soil particles were measured in the aggrading sites of the reservoir followed by the valley landform (south direction) and in the forested and afforested land systems of the catchment. Unfortunately, no previous study compared the influence of slope classes and elevation classes yet.
Taking into account the nature of texture, more clayey soil in the downstream part of the study site is expected to drain more water which permits longer growing periods to plants [15
]. The upstream observed with relatively higher sand particles would have higher infiltration and less water holding capacity. This could happen due to the presence of soil with larger pores in sandy particles. The finding of this research happens in a situation where the topography is rugged (Figure 2
) and the livelihood of people depends on subsistence food crop production.
Since the study watershed represents the agro-ecological feature of the northwestern highlands of the country, the statistical finding of the study could be useful information for the future related studies to be conducted in the larger tropical mountainous region. The observed result implies the presence of cultivated land use in steep slopes and siltation in the downstream. Unless corrective measures are taken in land use practices, in the long run, the existing situation could affect adversely the sustainability of crop land use and irrigation dam that may be constructed in the site.