Assessing Soil Properties and Landforms in the Mai-Negus Catchment, Northern Ethiopia

Pedosphere 26(5): 745–759, 2016 doi:10.1016/S1002-0160(15)60085-6 ISSN 1002-0160/CN 32-1315/P c 2016 Soil Science Society of China ⃝ Published by Elsevier B.V. and Science Press

Assessing Soil Properties and Landforms in the Mai-Negus Catchment, Northern Ethiopia Gebreyesus Brhane TESFAHUNEGN1,2,∗ , Lulseged TAMENE3 and Paul L. G. VLEK2 1 College

of Agriculture, Aksum University-Shire Campus, P.O. Box 314, Shire (Ethiopia) for Development Research (ZEF), University of Bonn, Walter-Flex-Str. 3, 53113 Bonn (Germany) 3 International Center for Tropical Agriculture (CIAT), Chitedze Agricultural Research Station, P.O. Box 158, Lilongwe (Malawi) 2 Center

(Received December 12, 2014; revised June 30, 2016)

ABSTRACT Soil degradation is a serious environmental problem in Ethiopia. However, little information is documented on indicators such as variations in soil properties across different landforms in a catchment. This study was aimed to assess soil properties and their changes across sites with different erosion statuses, and identify landscape positions that require prior management attention in the Mai-Negus catchment, northern Ethiopia. Three types of erosion-status sites (stable, eroding and aggrading) were identified using reconnaissance surveys, and then the corresponding soil samples were collected and analyzed. The major soil properties were significantly varied (P ≤ 0.05) among the three erosion-status sites. The highest soil pH, organic carbon, total nitrogen, cation exchange capacity, iron and zinc were recorded from the aggrading sites in the reservoir and valley landforms of the study catchment. A higher bulk density was generally recorded in the eroding sites, whereas a lower value was observed in the aggrading sites. The highest sand content was observed in the eroding sites of the mountain followed by the central ridge landform. The paired mean difference and the correlation matrix of most soil properties between the different erosion statuses also showed significant differences. About 95% of the erosionstatus sites were correctly classified by the discriminant function, indicating that the field survey-based classification was acceptable for decision making. On the basis of this study, suitable interventions should thus be introduced to the prioritized landforms, which are the mountain and central ridge, and eroding sites with severely degraded soil properties across the catchment. Key Words:

aggrading site, discriminant function, eroding site, erosion status, landscape, soil property variability, stable site

Citation: Tesfahunegn G B, Tamene L, Vlek P L G. 2016. Assessing soil properties and landforms in the Mai-Negus catchment, northern Ethiopia. Pedosphere. 26(5): 745–759.

Land degradation in terms of soil fertility or nutrient deterioration is a major threat to food security and natural resource conservation in Sub-Saharan Africa (Bationo et al., 2007). Soil erosion is the most serious cause of such environmental degradation, which constrains agricultural production and food security (UNEP and UNESCO, 1980; Eswaran et al., 2001; Haregeweyn et al., 2008). In relation to this, various studies indicated that Ethiopia is one of the countries most severely affected by erosion in Sub-Saharan Africa (FAO, 1986; El-Swaify and Hurni, 1996). Out of the 60 million ha of agricultural productive areas in Ethiopia, nearly 27 million ha have experienced erosion, 14 million ha have been severely eroded and requiring rehabilitation, and 2 million ha have lost top soil with an average soil loss of about 100 t ha−1 year−1 from cultivated lands (FAO, 1986; Mekonen and Tesfahunegn, 2011). Erosion is especially severe in the Tigray region of the northern Ethiopia, with an average soil loss from cultivated land of > 49 t ha−1 year−1 ∗ Corresponding

author. E-mail: [email protected].

(Tamene, 2005), exceeding the average soil loss of 42 t ha−1 year−1 for the whole of Ethiopia (Hurni, 1993). Soil degradation by water erosion is a major constraint to the sustainability of agricultural production since it influences several soil properties such as topsoil depth, soil organic carbon, nutrient status, soil texture, soil structure, water-holding capacity and water transmission characteristics (Haileslassie et al., 2005; Zougmor´e et al., 2010). For Ethiopia, Stoorvogel and Smaling (1990) predicted negative national nutrient balances due to erosion: −47 kg ha−1 for nitrogen (N), −15 kg ha−1 for phosphorus (P2 O5 ) and −38 kg ha−1 for potassium (K2 O), while inflows from fertilizers are very low (< 10 kg ha−1 ). This prediction is twice as high as the average nutrient depletion values for Sub-Saharan Africa, indicating a severe problem in Ethiopia. Haileslassie et al. (2005) also estimated soil nutrient depletion rates of 122 kg N ha−1 year−1 , 13 kg P2 O5 ha−1 year−1 and 82 kg K2 O ha−1 year−1 in Ethiopia. These studies indicated that soil erosion is

G. B. TESFAHUNEGN et al.

746

the major cause of such soil nutrient depletion, but its impact significantly varied among different landscape areas (Haileslassie et al., 2005). Nevertheless, degradation by soil erosion is often assessed in terms of loss of soil mass using erosion models rather than variability in soil nutrients and other soil properties (Pierce and Lal, 1994; Haregeweyn et al., 2008). In addition, little information is available on the spatial variability of soil physical properties and fertility/nutrients across different landforms at different levels of soil erosion. However, such information is crucial for developing soil management strategy targeted to hotspots of soil degradation (Pennock, 1998; Lal, 1999). Studies elsewhere have shown that erosion processes contribute significantly to the variability of soil properties and the associated nutrients within complex topography (e.g., Stone et al., 1985; Kreznor et al., 1989). Soil erosion and sediment delivery processes, which export sediment-bound nutrients to deposition areas, are influenced by landscape characteristics such as terrain, vegetation cover and soil erodibility (Withers and Lord, 2002; Muchena et al., 2005; Haregeweyn et al., 2008). Such soil depletion could be a fundamental cause for declining food production (Haileslassie et al., 2005; Haregeweyn et al., 2008). Furthermore, assessing soil property variability in relation to site characteristics such as landform units and soil erosion statuses is critical for predicting rates of ecosystem processes (Schimel et al., 1991) and understanding how a particular ecosystem works (Schimel et al., 1991; Kosmas et al., 2000). Despite of these facts, little is also understood and documented about the link between soil nutrient losses and landforms in Ethiopian catchments. The objective of this study was to bridge such research gaps by assessing

variability in soil properties at sites with different erosion statuses and identifying landforms that require management attention in the Mai-Neguse catchment of northern Ethiopia. MATERIALS AND METHODS Study area This study was conducted in the Mai-Negus catchment (38◦ 37′ 0′′ –38◦ 41′ 0′′ E and 14◦ 7′ 0′′ –14◦ 9′ 30′′ N) of the Tigray region, northern Ethiopia (Fig. 1a). The catchment has an area of 1 240 ha and altitudes ranging from 2 060 to 2 650 m above sea level (Fig. 1b). Mean annual temperature in the study area is 22 ◦ C and Mean annual precipitation is 700 mm. Most rainfall (> 70%) occurs between July and August (Ethiopian Meteorology Agency, Mekelle Branch). Cultivated land dominates over the other land-use types, with teff (Eragrostis tef) being the major crop in the study area. There are pastures and scattered patches of trees, bushes and shrubs. The catchment is dominated by lava pyroclastic and meta-volcanic rocks. Soils are mainly Leptosols on the steep slopes, Cambisols on the relatively gentle slopes and Vertisols on the flat areas (FAO, 1998). Landscape assessment Reconnaissance surveys were executed to get an overview about the general catchment characteristics and erosion statuses. Such data were collected from June to December 2009. Field observed data, information from a topographic map and preliminary report by Ethiopia Ministry of Water Resources (EMWR, 2002) were used to classify the catchment into different landforms (Fig. 1a). Considering elevation, slope and geo-

Fig. 1 Representative soil sampling points in different erosion-status sites in the Mai-Negus catchment of Tigray, Ethiopia (a) and the spatial distribution of elevation in the study catchment (b). a.s.l. = above sea level.

ASSESSING SOIL PROPERTIES AND LANDFORMS

747

morphic processes (e.g., surface flows and alluvial/ colluvial deposition), coordinates were collected and then the catchment topography was mainly classified into six landforms in ArcGIS (Fig. 1a). The landforms are the valley (19% of the catchment area), plateau (8%), rolling hill (9%), central ridge (27%), escarpment (29%) and mountain (6%) with an average slope of 4%, 13%, 18%, 22%, 36% and 80%, respectively. The reservoir was considered a separate landform because of its unique erosion and deposition processes at the toe-slope of the catchment. This landform covered about 1.2% of the catchment area in June 2009. The sediment deposited in the reservoir was used to assess the soil properties transported from the entire catchment. Deposition sites in the other landforms of the catchment were used to assess soils deposited on the way to the reservoir (outlet) after being transported from the original place. Generally, the field survey used geomorphological indicators such as thickness of alluvial/colluvial deposits and degree of truncation of top soil horizon when identifying the three erosion statuses, namely, stable, eroding and aggrading sites in the landforms of the study catchment (Fig. 1a). In the stable or less eroded sites, slopes are flat to gentle and there is little evidence of soil truncation or deposition, indicating that soil loss and gain are more or less balanced. A combination of the erosion indicators and features associated with soil profile thickness were used in identifying the eroding sites. Sediment deposition areas, such as depressions and flat areas that received sediment from the upper slopes through erosion channels, were considered and selected as the aggra-

ding sites. Selection of erosion-status sites and soil sampling points Different erosion-status sites (stable, eroding and aggrading) and the soil sampling points in each site were selected in four steps. First, an overview of landuses, lithology, soil types, slopes and elevation ranges of the landforms was documented after a reconnaissance survey (Table I). This was followed by informal discussions with farmers and development agents in the study area to gain insight about land-use history as well as land- and crop-management practices. Subsequently, three erosion statuses were selected according to soil morphological (soil profile) and erosion indicators (rills, gullies, surface sheet wash, exposure of roots and stones and depositions). Finally, soil/sediment sampling points that represented the erosion-status sites were located and geo-referenced (Fig. 1a). In this study, composite sampling was used because it is the most economical method. As a result, composite soil samples were collected from randomly selected points in the sampling plots to provide representative values. In each erosion-status site, two soil sampling points were identified. The area of the soil sampling points ranged from 150 to 300 m2 . Such areas were located considering the homogeneity in hydrological condition of each site on the basis of the researchers’ field observations and judgments. Based on the sampling plot area, 5–8 composite samples at 0–20 cm soil depth were collected from each sampling point. The soil samples collected from each sampling point were

TABLE I Biophysical description of the landforms in the Mai-Negus catchment, northern Ethiopia Landform

Areaa)

Land-use coverb)

Lithologyc)

Arable

Grazing

Bush and woode)

80 36 70 65 47 47 n.a.

10 34 13 31 16 31 n.a.

4 26 5 2 30 14 n.a.

Othersf)

BM

LP

ST

6 4 12 2 7 5 n.a.

100 35 5 77 58 84 100

n.a.h) 65 95 23 42 7 n.a.

n.a. n.a. n.a. n.a. n.a. 9 n.a.

% Rolling hill Mountain Central ridge Valley Plateau Escarpment Reservoir a) Total

10.0 14.5 25.5 19.9 9.8 19.1 1.2

Sloped)

Elevationd)

◦

m a.s.l.g) 2 150–2 240 2 350–2 650 2 230–2 450 2 070–2 100 2 500–2 550 2 270–2 540 2 060–2 080

3–16 4–79 3–25 0–6 3–10 3–30 0

catchment area is 12.40 km2 and total reservoir area is 0.15 km2 . proportion of land-use cover was derived from Landsat image of November 2009 overlaid by the landform map. c) This was derived by overlaying the landform map over the geology map: BM = basic metavolcanics; LP = lava pyroclastic; ST = sandstone. d) Developed from digital elevation model (DEM). e) Including closed area, plantation and natural vegetation. f) Including settlement, rock-out crop, marginalized area, etc. g) Above sea level. h) Not applicable. b) The

G. B. TESFAHUNEGN et al.

748

air-dried, homogenized and sieved to pass through a 2-mm sieve. A total of 36 soil samples (6 landforms × 3 erosion statuses × 2 sampling points = 36) and 6 sediment sampling points from the reservoir resulted in a grand total of 42 samples for laboratory analysis. Selection of sediment sampling points in the reservoir considered the flow direction from the source areas, outflow and duration where water was in the reservoir. The representative sampling points identified in the reservoir were located at the inlet in the northeast direction (the sampling point I), the inlet in the northwest direction (the sampling point II), the inlet in the north direction with sides and ridges around it (the sampling point III), the center part of the reservoir (the sampling point IV), the part near the reservoir spillway (the sampling point V), and the outlet of the reservoir (the sampling point VI). The spatial distribution of the sampling points in the reservoir and other 6 landforms is shown in Fig. 1a. There was no water at the selected sampling points of the reservoir in the 2nd week of June 2009 as rainfall was low during the 2008 rainy season. In each of the sampling points identified in the reservoir, 3 pits were opened and 3 composite samples were collected to represent the entire depth of each pit. The samples collected from each sampling point were pooled and mixed thoroughly in a basket and a sub-sample of 500 g was taken for analysis. Soil/sediment analysis Soil/sediment samples were analyzed for soil texture using the Bouyoucos hydrometer method (Gee and Bauder, 1986), bulk density (BD) by the core method (Blake and Hartge, 1986), electrical conductivity (EC) by the saturated paste extract (Rhoades, 1982), and pH at 1:2.5 sample to water ratio using a pH meter (Thomas, 1996), organic carbon (OC) by the Walkley-Black method (Bremmer and Mulvaney, 1982), available phosphorus (Pav ) by the Olsen method (Olsen and Sommers, 1982) and total nitrogen (TN) and total phosphorus (TP) by the Kjeldhal digestion method (Anderson and Ingram, 1989). Cation exchange capacity (CEC) was determined by the ammonium acetate extraction buffered at pH 7 (Rhoades, 1982). Exchangeable bases such as calcium (Ca), magnesium (Mg), potassium (K) and sodium (Na) were analyzed after the samples were extracted using 1 mol L−1 ammonium acetate at pH 7.0. Readings for Ca and Mg in the extracts were determined using an atomic absorption spectrophotometer, while K and Na were determined by the flame photometry (Black et al., 1965). Iron (Fe) and zinc (Zn) were determined by the method described in Baruah and Barthakur (1999)

using 0.005 mol L−1 diethylene triamine pentaacetic acid extraction. Soil porosity (%) was computed from the BD and average particle density (PD) of 2.65 Mg m−3 as the equation: soil porosity = (1 – BD/PD) × 100 (Brady and Weil, 2002). Exchangeable sodium percentage (ESP) was calculated by dividing exchangeable Na (ExNa) by CEC. Base saturation percentage (BSP) was calculated by dividing the sum of basic cations by CEC, multiplied by 100 (Coyne and Thompson, 2006). Assessing the link of soil properties among different erosion-status sites Based on the results of the soil/sediment analysis, the degree of association/relationship between soil properties among stable, eroding and aggrading erosionstatus sites in the landforms including the reservoir was assessed. This was carried out using the paired mean comparison and bi-variate correlation matrixes. Data analysis Data were subjected to descriptive analysis and analysis of variance (ANOVA) using SPSS 18.0 release software (SPSS Inc., 2011). One-way ANOVA was performed to test variations in means of soil properties across the different erosion-status sites at catchment level and with respect to the landforms. Soil datasets were first tested for normal distribution using Kolmogorov-Simirnov test statistics and those failing the test were transformed to normality. Differences between means were tested by the least significance differences (LSD) method at P ≤ 0.05. One-sample t-test was used to test soil property variability across the sampling points in the reservoir. Paired-samples student’s t-test at P ≤ 0.05 was also used for paired mean comparison between values of the same soil property in different erosion-status sites (stable vs. eroding, stable vs. aggrading and eroding vs. aggrading) of each landform. Selected properties were evaluated from the context of critical levels set by other studies. Correlations among the soil properties observed from different erosion statuses were checked by the Pearson’s productmoment correlation test (2-tailed test) and the significance levels were reported at P ≤ 0.05. In addition, multiple discriminant analysis was applied using SPSS software package to determine whether the soil properties measured in standard soil survey could be used to correctly classify the erosion statuses. Given a set of independent variables the discriminant analysis is used to determine linear combinations of those variables that best discriminate between a priori-defined group of dependent variables (e.g., stable, eroding and aggrading) by maximizing the between-group variance relative to the within-gro-

ASSESSING SOIL PROPERTIES AND LANDFORMS

749

up variance (Everitt and Dunn, 1992; Hair et al., 1992; Tesfahunegn et al., 2011). These linear combinations are called discriminant functions, which are described as the equation: fkm = u0 + u1 X1 km + u2 X2km + · · · + up Xpkm

(1)

where fkm is the value (score) of discriminant function for Case m (sampling point) in Group k (erosion status), Xpkm is the pth discriminant variable (soil property) for Case m in Group k, and up is the pth standardized coefficient. In the discriminant analysis, the step-wise method was applied to choose variables for the equation on the basis of how much they lower Wilks’ lambda. Smaller values of Wilks’ lambda indicate greater discriminatory ability of the function (SPSS Inc., 2011). A variable entered into the model if the significance level of the F -test is less than 0.05 and removed if the value is greater than 0.10. The procedure automatically selected the first function (Function 1) that was able to separate the groups as much as possible; a second function (Function 2) was then selected that was both uncorrelated with the first function and that provided as much further separation as possible. The number of functions is equalled to the number of group variables (dependent) minus one (SPSS Inc., 2011). Discriminant functions were interpreted by means of standardized coefficients. The larger absolute values of standardized coefficients before varialbles corresponded to the variables with greater discrimination ability between the groups of erosion statuses (Everitt and Dunn, 1992). Discriminant scores were computed for each case to determine its position along the axis. Scores of cases in each group variable are averaged and then referred to as the group centroid. Examining the score means for different functions across groups allow one to determine which group is discriminated by discriminant function. The classification functions in the discriminant analysis were also used to determine group cases that most likely belonged to the original and cross-validation cases using a classification rate. In cross validation, each case is classified by the function derived from all cases other than that case alone (SPSS Inc., 2011). RESULTS AND DISCUSSION Soil property variability across different erosion-statuse sites at catchment level There were significant differences (P ≤ 0.05) in the means of soil physical properties among the erosion-

status sites at catchment level (Table II). For instance, significant differences in soil BD and porosity between the eroding and stable, and eroding and aggrading sites were observed, but these properties had no statistical (P > 0.05) differences between the stable and aggrading sites (Table II). Generally, the mean BD was higher in the eroding sites (1.66 Mg m−3 ) followed by the stable sites (1.39 Mg m−3 ), whereas a lower BD was recorded in the aggrading sites (1.34 Mg m−3 ). The BD in the eroding sites of the catchment showed above the critical level of 1.00–1.40 Mg m−3 as an ideal condition for root growth reported by Arshad et al. (1996). The mean clay content in the aggrading sites was significantly higher than those in the eroding and stable sites (Table II). This could be attributed to the mobilization of clay-rich soils from the source areas (eroding sites) to the aggrading sites. However, the clay content was not significantly different between the stable and eroding sites. Such non-significant difference could be resulted from anthropogenic factors that increase clay content in the eroding sites by mixed-up of clay-rich sub-soils. In line with the present finding, previous studies (e.g., Cruickshank, 1972; Tesfahunegn, 2010) reported that clay content increased with depth consistently from the surface horizon to the sub-surface horizon of a profile, while sand decreased with depth. The reason for the high clay content with depth is attributed to the process of translocation of clay from the upper to the lower horizons (Cruickshank, 1972; Tesfahunegn, 2010). The highest sand (51%) was observed in the eroding sites, whereas the lowest was in the aggrading sites (27%). The highest silt was observed in the aggrading sites. Generally, the soil texture in the stable sites was classified as loamy soil, whereas in the eroding and aggrading sites as sandy clay loam and clay loam, respectively (Table II). Such differences in soil texture indicated that erosion removes surface soil particles which are rich in organic materials, exposes a lower soil layers and leaves coarser particles in the eroding sites. In such conditions, water retention capacity of eroded soils decreases, but if the lower soil horizons have greater clay contents due to natural and human factors than the removed surface layers initially, soil water retention capacity may increase (Cruickshank, 1972). Generally, this study confirms the basic principle that silt and clay are the soil components susceptible to erosion selective detachment and transport processes and accumulate in depositional sites (Brady and Weil, 2002). Soil nutrients adsorbed to the finer particles are also exportedto reservoirs and gorges (Haregeweyn et al., 2008), in which such areas are not useful for agricultural production. Intervention that

G. B. TESFAHUNEGN et al.

750

TABLE II Means of soil properties (at catchment level) determined from different erosion-status sites in the Mai-Negus catchment, northern Ethiopia Soil propertya)

Sand (%) Silt (%) Clay (%) Soil texture BD (Mg m−3 ) Porosity (%) pH EC (dS m−1 ) OC (g kg−1 ) Pav (mg kg−1 ) TN (g kg−1 ) TP (mg kg−1 ) ExK (cmolc kg−1 ) ExCa (cmolc kg−1 ) ExMg (cmolc kg−1 ) ExNa (cmolc kg−1 ) Sum of basic cations (cmolc kg−1 ) CEC (cmolc kg−1 ) BSP ESP Fe (mg kg−1 ) Zn (mg kg−1 )

Critical levelc)

Erosion-status site Stable (n = 12)

Eroding (n = 12)

Aggradingb) (n = 18)

42.0bd) 37.8b 20.2b Loam 1.39b 46.7a 6.91b 0.18b 22.4a 12.52b 2.0b 946b 0.93a 18.3b 12.0a 0.33b 31.5a 36.0b 87.0a 0.90b 23.9b 8.5b

51.3a 25.7c 23.0b Sandy clay loam 1.66a 23.7b 6.33c 0.15b 4.5b 2.50c 0.9c 363c 0.59b 10.1c 4.6b 0.29b 17.6b 21.0c 78.9b 1.30a 6.0c 1.32c

26.7c 43.2a 30.1a Clay loam 1.34b 51.3a 7.56a 0.29a 25.8a 17.63a 5.8a 1571a 0.96a 25.6a 11.5a 0.44a 38.5a 45.7a 83.9a 0.96b 29.4a 11.8a

1.00–1.40e) 6.5–8.5 5–10.0 10–15 < 10.0 < 200 < 0.2 < 5.0 < 1.5

< 50.0 < 20.0

a) BD

= bulk density; EC = electrical conductivity; OC = organic carbon; Pav = available phosphorous; TN = total nitrogen; TP = total phosphorous; ExK = exchangeable potassium; ExCa = exchangeable calcium; ExMg = exchangeable magnesium; ExNa = exchangeable sodium; CEC = cation exchange capacity; BSP = base saturation percentage; ESP = exchangeable sodium percentage. b) Synonymous with ‘deposition’ in the context of this study. c) Ulrich (1947) defined critical level as the narrow range of concentration at which growth rate or yield can decline in comparison with a higher nutrient level. Critical levels for micronutrients are sourced from Jones et al. (1973), and those for macronutrients are from S´ anchez et al. (1982). d) Means followed by different letters in the same row are significantly different at P ≤ 0.05. e) According to Arshad et al. (1996), the ideal soil BD for root growth is < 1.40 Mg m−3 .

improves soil organic matter (SOM) and maintains the soils in the eroding sites is thus crucial to restore such soils and achieve a sustainable agricultural production. In this study, soil Pav , TN, TP, pH, exchangeable Ca (ExCa), CEC, Fe, and Zn showed significant differences among the stable, eroding and aggrading sites at catchment level (Table II). Statistically significant differences in EC and ExNa were also observed between the aggrading and stable, and aggrading and eroding sites. However, these soil properties did not show significant differences between the stable and eroding sites. There were significant differences in OC, exchangeable K (ExK), exchangeable Mg (ExMg), sum of basic cations and BSP between the aggrading and eroding sites, but these properties did not show significant differences between the aggrading and stable sites at catchment level (Table II). Despite of this, lower values of the mean soil chemical properties were observed in the eroding sites. This is confirmed by the finding reported in Haregeweyn et al. (2008): erosion is the main

factor for soil nutrient depletion that ranges from 63% to 80% of the total nutrient losses in Ethiopia. Generally, the analytical results of soil properties in this study have clearly indicated that the soils in the eroding sites need efforts that geared not only to improve soil system but also to improve water use efficiency and soil water content. The report of Tesfahunegn (2010) also showed that technologies that improve both soil fertility and water availability are equally demanded in degraded soil conditions in semi-arid Ethiopia. In addition, this study showed that mean soil pH, OC, TN and Pav in the eroding sites of the catchment were found below their lower critical levels required for normal plant growth. But the values of TP, ExCa, ExMg and ExK showed above their critical levels in all the erosion-status sites. This might be attributed to their high stock in the source soil profiles. The soil nutrients such as TN, Fe and Zn were poor in all the erosion-status sites in the study catchment even though the lowest was observed in the eroding sites

ASSESSING SOIL PROPERTIES AND LANDFORMS

751

(Table II). The mean soil pH (Table II) showed slightly acidic in the eroding sites, whereas moderately alkaline in the aggrading sites. This can imply that basic cations and organic matter are removed by erosion from the eroding to aggrading sites. Consistent with the present finding of higher pH in the aggrading sites followed by the stable sites, previous studies elsewhere (e.g., Battison et al., 1987; Pennock et al., 1994; Papiernik et al., 2005, 2007) reported that soil pH was high in sites with fertile soils such as the aggrading areas due to higher SOM content and vice versa in the eroding sites. This study generalized that the highest means of soil chemical properties were found in the aggrading sites and the lowest in the eroding sites of the catchment (Table II). The finding was consistent with the results of some studies (Gregorich and Anderson, 1985; Heckrath et al., 2005; Papiernik et al., 2005, 2007) that soil chemical properties such as OC, TN and Pav were higher in areas with large mass of soil accumulated (deposited) as compared to the source areas where soil has been removed by erosion. The report by Thomas and Cassel (1979) also confirmed that soil erosion in the eroding sites leads to deficiency in nutrients

such as TN, OC and Zn. In such conditions, literature suggests that implementation of suitable soil management practices, e.g., soil-landscape rehabilitation (moving soils from the deposited to topsoil-depleted areas), and land-use practices that improve soil systems are too crucial considering soil property variability across the erosion statuses (Legendre et al., 2002; Dale and Fortin, 2005; Papiernik et al., 2005, 2009). Soil property variability across the landforms in each erosion status In stable sites. The statistical analyses of selected soil properties determined from the stable sites across the landforms in the Mai-Negus catchment are presented in Table III. The results showed that most soil properties were significantly (P ≤ 0.05) different across the landforms. However, there were nonsignificant differences in soil porosity, TP, ExMg and Zn among different landforms. In addition, most soil properties did not show significant differences between the mountain and central ridge landforms (Table III). In the stable sites across the landforms the highest soil sand content (52%) was observed in the mountain followed by the central ridge (46%), whereas the lowest

TABLE III Means of soil propertiesa) with respect to stable, eroding and aggrading sites across the landforms in the Mai-Negus catchment, northern Ethiopia LFb) Sand Silt Clay BD %

1 2 3 4 5 6

43bcc) 52a 46b 42c 32d 43bc

38c 38c 32d 49a 32d 43b

1 2 3 4 5 6

47c 69a 63ab 58b 37d 50c

1 2 3 4 5 6 7

28b 46a 45a 14d 16d 16d 22c

a) See

Mg m−3 19bc 10cd 22b 9d 36a 14cd

Poro- pH sity

EC

%

dS m−1 g kg−1

1.47a 48a 1.39b 42a 1.42b 43a 1.33c 50a 1.36bc 49a 1.41b 48a

6.98a 0.13b 6.99a 0.15b 6.85ab 0.11c 7.20a 0.18a 6.43b 0.13b 6.99a 0.14b

29a 24b 27a 4c 21b 16c 25ab 17b 25ab 38a 29a 22b

1.66c 24a 1.69b 25a 1.72a 24a 1.58e 25a 1.63d 22a 1.65cd 21a

6.31b 0.08b 6.11b 0.13a 6.09b 0.09b 6.77a 0.15a 6.28b 0.09b 6.43ab 0.14a

41c 47b 36d 55a 35d 46b 53a

1.37b 1.46a 1.43a 1.26c 1.38b 1.37b 1.13d

7.35b 7.18b 7.01b 7.43b 7.30b 7.22b 8.04a

31c 7e 19d 31c 49a 38b 25cd

48c 50c 47c 56b 48c 47c 60a

0.14b 0.11b 0.14b 0.69a 0.17b 0.22b 0.55a

OC

9.4b 5.7c 6.6c 18.6a 10.0b 9.5b 4.6bc 2.4d 2.7cd 8.1a 4.9b 4.7bc 10.5d 6.2e 7.1e 24.4b 11.8c 11.0cd 36.5a

Pav

TN

TP

ExK

mg g mg kg−1 kg−1 kg−1 Stable site 13.2b 1.38c 1 014a 0.73c 9.80c 1.37c 879ab 0.64c 9.40c 1.78b 995a 0.62c 14.2ab 2.10a 907a 1.38a 16.1a 1.43c 945a 0.80c 12.4b 1.26d 935a 1.10b Eroding site 6.73a 0.96ab 605a 0.53c 5.01a 0.75b 530a 0.38d 4.45a 0.90ab 511a 0.36d 7.02a 1.15a 561a 0.67b 9.04a 1.02a 678a 0.57c 6.90a 0.81ab 648a 0.75a Aggrading site 14.8bc 1.74cd 1 439b 0.83c 12.2c 1.59d 1 372b 0.79c 12.3c 2.12c 1 343b 0.76c 23.0a 3.11b 1 626b 1.42a 21.4a 1.85c 1 512b 0.93bc 15.6b 1.66d 1 497b 1.31ab 24.9a 8.30a 2 206a 1.45a

ExCa

ExMg ExNa CEC Fe

cmolc kg−1

Zn mg kg−1

17.6b 10.9c 13.5c 23.5a 20.1ab 18.0b

11.8a 10.7a 11.8a 13.3a 10.4a 11.5a

0.26b 0.25b 0.27b 0.34a 0.35a 0.32a

34b 29c 28c 46a 38b 39b

20c 19c 17c 34a 27b 25b

8.04a 8.08a 7.50a 9.54a 8.75a 9.21a

9.2bc 5.5d 7.6cd 12.7a 14.0a 11.9ab

6.61a 5.24a 5.23a 6.25a 5.88a 6.50a

0.23c 0.17d 0.23c 0.34a 0.30b 0.28b

19b 13c 14c 21a 22a 21a

12.4a 10.1a 8.60a 19.8a 18.8a 16.6a

4.2ab 3.7b 3.b 4.4ab 5.5a 5.6a

18.3cd 13.0e 17.1d 29.0a 25.9b 21.6c 26.3ab

13.2b 12.6b 12.9b 15.8a 12.2b 13.9ab 14.0ab

0.31b 37c 0.28c 31d 0.29bc 33d 0.65a 48a 0.26cd 44b 0.30bc 43b 0.63a 49a

22c 20c 17cd 46a 31b 28b 44a

8.6c 8.2bc 7.9c 12.7ab 9.7bc 10.5bc 13.8a

Table II for descriptions of soil properties. is landform: 1 = rolling hill; 2 = mountain; 3 = central ridge; 4 = valley; 5 = plateau; 6 = escarpment; 7 = reservoir. c) Means followed by different letters in the same column are significantly different for the same erosion-status site at P ≤ 0.05. b) LF

752

(32%) in the plateau. Higher soil silt (49%) was observed in the stable sites of the valley as compared to the other landforms, implying that fine soil materials are transported by erosion from the mountain or central ridge to flat areas such as the valley landform. The lowest soil BD (1.33 Mg m−3 ) was observed in the stable sites of the valley, whereas the highest (1.47 Mg m−3 ) in the central ridge (Table III). The soil BD in the stable sites of the central ridge was slightly above the critical level for root growth, which was reported by Arshad et al. (1996). Higher soil pH, OC, TN, CEC, Fe, Zn and exchangeable cations were observed in the stable sites of the valley. A higher soil Pav was observed in the stable sites of the plateau, which might be due to little losses of P and/or its high mineralization. This study generalized that the lowest soil nutrients were observed in the stable sites of the mountain and central ridge landforms (Table III). As a result, soil Pav content in these landforms was within the range of the critical level, while in the other landforms it was found above the critical level for plant growth, which was reported by S´anchez et al. (1982). In the stable sites across all landforms, soil TN, Fe and Zn were below the critical levels required for normal plant root growth, even though the lowest values were found in the mountain and central ridge landforms. This could be attributed to the fact that the soils in these sites are deteriorated through time with removals such as crop and grass harvests and the losses related to leaching but with fewer turnovers from vegetation by-products and other inputs. Plaster (1992) also reported that plant consumed more soil N than any other nutrient. Zn deficiency is commonly observed in crops, animals and humans in developing countries such as Ethiopia (Fageria, 2009), which could confirm the reason for Zn to be below its critical level in this study. The same author also reported that deficiencies of micronutrients (i.e., Fe) were often induced because of interaction effects with other macronutrients. In eroding sites. There were significant differences in soil sand, clay, silt, BD, pH, EC, OC, TN, ExK, ExCa, CEC and Zn in the eroding sites across the landforms (Table III). The highest sand (69%) was observed in the eroding sites of the mountain, which was higher than those in the stable and aggrading sites. Such high sand content could be attributed mainly to erosion that detached and transported fine soil materials, leaving coarser particles in the source soil (Brady and Weil, 2002). Higher clay was observed in the eroding sites of the plateau followed by the valley. Regardless of the non-significant differences in soil porosity,

G. B. TESFAHUNEGN et al.

Pav , TP and Fe across the landforms, higher values of these soil properties were found in the valley (Table III). The non-significant differences indicated that the degree of degradation is almost similar. The highest values of most chemical soil properties and fine soil materials were found in the eroding sites of the valley, whereas the lowest values in the mountain (Table III). Nevertheless, a higher soil BD was observed in the eroding sites of central ridge (1.72 Mg m−3 ) as compared to the lowest value in the valley (1.58 Mg m−3 ), indicating that there was a difference in the effect of erosion across the landforms. The soil BD in the eroding sites of all landforms was higher than the critical level determined by Arshad et al. (1996). The soil pH, OC and ExMg in all landforms except the valley and soil TN, Pav , Fe and Zn in all landforms (Table III) were below the critical levels for plant growth reported by Jones et al. (1973). Such results indicated that attention should be given to improve soil properties above or within the critical level in the eroding sites across the landforms. In aggrading sites. There were significant differences in most soil properties determined in the aggrading sites across the landforms (Table III). The lowest sand content was observed in the aggrading sites of the valley (14%) followed by the reservoir (22%), whereas the highest sand (41%) was found in the mountain and central ridge landforms. Higher silt content was observed in the aggrading sites of the valley and reservoir landforms (Table III). The lowest soil BD was recorded in the reservoir (1.13 Mg m−3 ) followed by the valley landform (1.26 Mg m−3 ). A higher soil BD was found in the aggrading sites of the mountain (1.46 Mg m−3 ) and central ridge (1.43 Mg m−3 ) landforms (Table III). This may be attributed to the very low soil OC and high weathering of rocky sub-soil remnants that are deposited in the aggrading sites of these landforms. The values of soil Pav , TN, OC, TP, ExK, Ca, Mg, CEC, Fe and Zn determined in the reservoir (the aggrading sites) were higher than those in the other landforms (Table III). The lowest soil nutrients were observed in the aggrading sites of the mountain and central ridge landforms and thus the most severely degraded soil properties were found in these two landforms. Such results indicated that soil nutrients associated with fine soil particles are transported by erosion mainly to the reservoir from the erosion sources in the mountain and central ridge landforms. Consistent with the results of this study, Kreznor et al. (1989) reported that landscape position and erosion levels influenced soil properties in Illinois watershed.

ASSESSING SOIL PROPERTIES AND LANDFORMS

753

Although soil TN, Fe and Zn were higher in the aggrading sites as compared to the stable and eroding sites in the landforms, these soil nutrients were below their critical levels for normal plant growth reported by Jones et al. (1973) and S´anchez et al. (1982). The other soil nutrient alone in the aggrading sites were above their critical levels. Despite of these facts, the report by Haregeweyn et al. (2008) showed that the level of individual nutrients alone may not reflect soil fertility status. These authors thus stated that the interaction and proportion between each soil chemical properties and also with other soil physical and biological properties are equally important.

The variability among the sampling points in the reservoir could be associated with the sources of sediment and activities in the upper stream of the catchment such as settlement, agriculture and area exclosure. With this view, field observation showed higher silt at the sampling point I, in which it was demonstrated that the soil texture of the source area was silt-dominated. In addition to the sediment sources, the mechanism of sediment settlement may influence the texture and depth of sediment deposited in the reservoir. However, this merits further investigation to understand how such mechanisms influence sediment depth, pattern and its characteristics in a reservoir. In line with this study, Saavedra (2005) reported that sediment depositions were largely in a non-uniform pattern in all the surveyed reservoirs. However, these results disagreed with the finding reported by Tamene (2005), who reported that there was a uniform pattern in sediment deposition in a given reservoir.

Soil property variability across the reservoir The one-sample t-test analysis of the soil properties determined from the reservoir sampling points showed significant differences (Table IV). For example, the sand (30%) deposited in the sampling point III (inlet north-direction and side-edges) of the reservoir was significantly higher than in the other sampling points. The highest silt (67%) was deposited in the sampling point I (inlet northeast-direction) and the highest clay (51%) in the sampling point II (inlet northwest direction). The trend of variability in the soil physical and chemical properties across the sampling points in the reservoir was given by their lower and upper confidence intervals (Table IV). The results of this study thus tested and accepted the alternative hypothesis, i.e., soil properties varied significantly across the sampling points in the reservoir.

Relationship of soil properties among the erosion statuses The degrees of association/relationships between soil properties in different erosion-status sites are presented in Table V. The paired mean comparision of the same soil property among the stable, eroding and aggrading erosion-status sites showed significant differences (P ≤ 0.05) for all paired soil properties, except the paired BD, ExK, ExMg and TP between the stable and aggrading sites (Table V). The soil property with non-significant and/or poor correlation of paired mean

TABLE IV Means of soil propertiesa) determined from 6 sediment sampling points in the reservoir in the Mai-Negus catchment, northern Ethiopia Item Mean 95% CIc) t valued) dfe) P value Item

Sand

Silt

15 ± 13.9–27.6 10.64 5 0.000

2.09b)

TN

TP

a) See

g 8.3 ± 0.4 6.8–10.0 7.94 5 0.000

25 ± 5.08 20.9–44.9 8.25 5 0.002 Fe

pH

EC

Mg 1.13 ± 0.01 1.08–1.20 49.67 5 0.000

Zn

2206 ± 173 1282–2371 21.55 5 0.001

mg 44 ± 4.92 38.4–53.3 22.3 5 0.000

OC m−1

8.05 ± 0.10 7.48–8.45 42.57 5 0.000

ExK

dS 0.55 ± 0.05 0.42–0.68 10.91 5 0.031

ExCa

ExMg

kg−1

g 36.5 ± 3.0 28.0–39.0 9.93 5 0.002

Pav mg kg−1 24.9 ± 1.19 21.1–27.7 24.04 5 0.005

ExNa

CEC

0.63 ± 0.04 0.53–0.73 16.31 5 0.000

49 ± 2.9 44.0–55.0 23.32 5 0.000

kg−1

kg−1

Table II for descriptions of soil properties. ± standard error (n = 6). c) 95% confidence interval. d) One-sample t-test value; e) Degree of freedom. b) Mean

BD m−3

% 60 ± 6.35 19.3–65.2 5.65 5 0.005

kg−1

Mean 95% CIc) t valued) dfe) P value

Clay

14 ± 1.2 12.6–15.7 12.7 5 0.000

1.27 ± 0.10 1.01–1.53 12.60 5 0.001

26 ± 1.24 21.1–29.5 21.3 5 0.000

cmolc 14 ± 0.80 13.0–16.1 25.48 5 0.001

G. B. TESFAHUNEGN et al.

754

TABLE V Paired mean difference (PMD) comparison and correlation tests for selected soil properties between erosion-status sites in the MaiNegus catchment, northern Ethiopia (n = 42) Paired soil propertya)

PMD

rb)

Paired soil property PMD

Sand (S vs. E) Sand (S vs. A) Sand (E vs. A) Silt (S vs. E) Silt (S vs. A) Silt (E vs. A) Clay (S vs. E) Clay (S vs. A) Clay (E vs. A) BD (S vs. E) BD (S vs. A) BD (E vs. A) pH (S vs. E) pH (S vs. A) pH (E vs. A)

−9.29** 15.28** 24.57** 12.05** −5.47** −17.5** −4.77* −9.82** −7.05** −0.16** −0.03ns 0.19** 0.27** −0.21* −0.48**

0.88** 0.19nsc) −0.18ns −0.84** 0.82** −0.92** −0.86** 0.78** −0.85** 0.92** 0.70* 0.88** 0.86** 0.48ns 0.76**

ExK (S vs. E) ExK (S vs. A) ExK (E vs. A) ExCa (E vs. A) ExCa (S vs. A) ExCa (E vs. A) ExMg (S vs. E) ExMg (S vs. A) ExMg (E vs. A) CEC (S vs. E) CEC (S vs. A) CEC (E vs. A) OC (S vs. E) OC (S vs. A) OC (E vs. A)

0.34** −0.03ns −0.37** 6.16** 2.06* −8.22** 7.36** 0.45ns −6.91** 13.91** −3.70* −17.61** 1.25** −0.23* −1.38**

r

Pair soil property PMD 0.92** 0.74** 0.83** 0.86** 0.80** 0.71** 0.58* 0.40ns 0.65* 0.87** 0.62* 0.89** 0.78** 0.83** 0.72**

TN (S vs. E) TN (S vs. A) TN (E vs. A) Pav (S vs. E) Pav (S vs. A) Pav (E vs. A) TP (S vs. E) TP (S vs. A) TP (E vs. A) Fe (S vs. E) Fe (S vs. A) Fe (E vs. A) Zn (S vs. E) Zn (S vs. A) Zn (E vs. A)

r

0.12** 0.80** −0.03** 0.83** −0.15** 0.64* 9.19** 0.64* −2.73* 0.67* −11.92** 0.84** 606.69* −0.33ns −1 406.65ns 0.69* −1 067.35* 0.83** −6.67** 0.93** −13.47* 0.55ns −6.68** 0.69* 9.39** 0.90** −0.22* 0.61* −9.62** 0.59*

*, **Significant at P ≤ 0.05 and P ≤ 0.01, respectively. a) See Table II for descriptions of soil properties. S vs. E = stable vs. eroding; S vs. A = stable vs. aggrading; E vs. A = eroding vs. aggrading. b) Correlation coefficient. c) Not significant.

difference between erosion statuses indicated that the source soil is not the determinant factor for soil composition in the aggrading sites; rather it could be other non-point sources of sediment within the landforms. Conversely, the significant paired mean difference of a particular soil property between erosion statuses indicated that soil erosion is the influential factor for their variability. The correlation coefficients of most soil properties between paired erosion statuses showed significant (P ≤ 0.05) differences (Table V). For example, the relationship of soil sand between the stable and eroding sites was significantly and positively correlated (r = 0.88, P = 0.01). This implied that if the sand in the stable sites was high, the possibility of getting higher sand in the eroding sites was explained by 77%. However, the negative and significant correlation of soil silt between the stable and eroding (r = −0.84), and eroding and aggrading (r = −0.92) sites showed an inverse relationships. The relationships of clay between the eroding and aggrading (r = −0.85), and stable and eroding sites (r = −0.86) were also significantly (P ≤ 0.05) correlated. Such relationships showed that the silt/clay decreased in the stable sites and increased in the eroding sites as described by the degree of correlation due to erosion effects. Similarly, the negatively strong correlations of silt/clay between the aggrading and eroding sites implied that their contents in the aggrading sites increased as the result of decreases in the eroding sites by the rate of their associa-

tion. Such strong correlations between these erosion statuses indicated that the silt and clay fractions in the aggrading sites are linked mainly to their source areas in the eroding sites. In addition, the results in this study indicated that the soil nutrients could be transported to the aggrading sites mainly attached with the fine soil particles as explained by their significantly higher correlation coefficients (Table VI). For instance, higher correlations between the fine soil particles and CEC, OC, Pav or TN were observed. The moderately and significantly negative correlation (r = −0.73, P = 0.01) between soil OC and BD explained the inverse association between the erosion statuses. The correlation matrix (Table VI) also showed statistically significant but negative correlations of sand and BD with most of the soil properties as expected. The soil TN in the aggrading sites was mainly organic origin as TN was moderately correlated with OC (r = 0.81, P = 0.01) and the remaining sources of TN could be inorganic fertilizers, weathering of parent materials and/or atmospheric deposition. The result of this study was consistent with the findings of Nye and Stephen (1962), who stated that SOM is an important reserve of soil TN. Soil CEC was significantly and positively correlated with the other soil properties determined from the erosion-status sites, except sand and BD. In this study, the correlation results showed that the influence of clay on CEC exceeded that of SOM and soil pH, since the correlation of CEC with clay was higher (r = 0.86)

ASSESSING SOIL PROPERTIES AND LANDFORMS

755

as compared with OC (r = 0.69) and pH (r = 0.43). In agreement with this result, previous studies reported that soil CEC is mainly influenced by clay minerals followed by SOM and soil reaction (pH) (Thomas, 1982; Landon, 1991). Micro nutrients such as Fe and Zn showed higher correlations with OC as compared to the other soil properties (Table VI), implying that their main source was organic matter. The soil Pav in the study catchment was also mainly organic origin as Pav was correlated with OC at r = 0.59, and the remaining could be from mineral sources such as inorganic fertilizers and weathering of parent materials. Such moderate to strong association of OC with most of the soil properties determined implied that the introduction of suitable interventions should be thoroughly focused on improving SOM in the catchment landforms. In line with the present study, Brierley et al. (2006) reported a strong link or connectivity or transfer of soil materials among different erosion statuses considering landscape configuration within a catchment. The report by Wan and E1-Swaify (1997) also showed that sediment transported by overland flow from the eroding to aggrading sites has long been recognized to be enriched with silt and clay particles, OC, soil nutrients and other soil chemicals when compared with the insitu soil. Many other studies reported that interrill erosion resulted in selective removal of silt and clay particles and the attached soil nutrients from the source areas to aggrading sites (e.g., Wan and E1-Swaify, 1997; Brierley et al., 2006). In addition, factors other than soil erosion such as removals with harvested products, crop residues, leaching and gaseous losses can reduce soil nutrients of the source soils. However, detailed investigation of the effects of such factors on soil nutrients at landform level in the study area merits further study. Multiple discriminant analysis Among the measured data of 21 soil properties 7 soil properties (silt, BD, CEC, Zn, TN, Pav and Fe) were selected by the step-wise method in the discriminant analysis. Factor analysis was also used to check the redundancy of variance (multicollinearity) of the datasets. The high factor loadings identified by the factor analysis (data not shown) were similar to the variables selected using step-wise statistics in the discriminant analysis. The standardized coefficients of discriminant function showed that silt, CEC, BD, Zn, TN, Pav and Fe were ranked from the most to the least important in discriminating between Group 1 (eroding) and the combination of Group 2 (stable)

G. B. TESFAHUNEGN et al.

756

and Group 3 (aggrading) categories (Function 1). The best discriminator in Function 2 was in the descending order of CEC, TN, Pav , Zn, Fe and silt (Table VII). In agreement with this finding, the other studies showed that CEC is identified as the best discriminator between poor and high soil quality in northern Ethiopia (Tesfahunegn et al., 2011) and among different soil groups (alluvial soil, colluvial soil and vertic soil) in Cambodia (Oberth¨ ur et al., 2000). The study of Shukla et al. (2006) using the discriminant coefficients also confirmed that BD and silt are among the most useful variables for assessing changes in soil quality under different land-uses and management systems in Ohio, USA. In general, about 84% of the variance in the discriminant model was explained due to Function 1. Function 1 discriminated between the eroding sites as one group and combination of the stable and aggrading sites as another group, and the remaining 16% by Function 2 which discriminated between the stable and aggrading sites. This showed that the variability was higher between the group of the eroding sites and the combined group of the stable and aggrading sites as compared to that between the stable and aggrading sites. In addition, the relationship of each dependent group variables (erosion statuses) with the independent variable (soil properties) is shown in Table VII. The unstandardized coefficients of discriminant function showed that Pav followed by silt and Zn in the eroding sites, Zn followed by TN and Pav in the stable sites, and BD followed by TN and CEC in the aggrading sites were the powerful predictor variables. The Zn for being among the high predictive power was confirmed by the report of Fageria (2009), who stated that soil Zn deficiency is the 5th leading risk factor for hu-

man beings in developing countries. In addition to the unstandard coefficients, the coefficients of determination (R2 ) of 0.99, 0.92 and 0.93 explained the effects of the 7 independent soil variables on group variables as stable, eroding and aggrading sites, respectively. Visualization of the functions in discriminating the group variables by plotting the individual scores of each case (sampling point) is presented in Fig. 2. It showed that Function 1 discriminated mainly between the group of eroding sites and the combined group of stable and aggrading sites because the cases of the eroding sites fell to the left of the centerline (0) but the combined group mainly fell near to the centerline towards the right of Function 1. In the vertical direction (Function 2) no cases of the eroding sites fell above the centerline (0). However, most of the cases in the stable sites fell above the centerline of Function 2, and majority of the cases for the aggrading sites fell below the centerline (0) of Function 2. This clearly demonstrated that Function 2 discriminated between the stable and aggrading sites (Fig. 2). This study also assessed how well the discriminant function correctly classified the cases of each group variable (Table VIII). The chance that the cases belong to the group variable was based on the values of the 7 soil properties identified by the step-wise method. The prior weighted cases used in the discriminant analysis for the stable, eroding and aggrading group variables were 12, 12, and 14, respectively. For the original grouped cases, the discriminant analysis correctly classified 11 of the 12 eroding cases and all the 12 stable cases, with a correct classification rate of 91.7% and 100%, respectively. In addition, the discriminant analysis correctly classified 13 of the 14 aggrading cases, with a correct classification rate of 92.9%. The discri-

TABLE VII Coefficients before selected soil propertiesa) (variables) and other parameters for the functions of multiple discriminant analysis Item

Coefficient

Constant

R2b)

P value

Silt

CEC

BD

Zn

TN

Pav

Fe

1.472 0.507

1.433 1.706

−1.326 0.843

1.197 −0.255

0.508 1.629

0.318 0.914

0.122 0.338

−5.275 −6.380

0.84 0.16

0.000** 0.001**

1.080 3.362 1.329

1.530 2.250 1.590

1.487 −2.218 2.247

2.422 −3.291 1.487

2.300 2.230 2.079

1.633 3.591 1.274

0.780 2.129 0.989

−7.731 −12.810 3.854

0.99 0.92 0.93

0.030* 0.005** 0.000**

Functionc) 1 2 Groupd) Stable Eroding Aggrading

*, **Significant at P ≤ 0.05 and P ≤ 0.01, respectively. Table II for descriptions of soil properties. b) Coefficient of determination. c) Wilks’ lambda test of functions was described by the standardized coefficients, showing the overall discrimination between groups. d) Described by the unstandardized coefficients using the functions for prediction within groups. a) See

ASSESSING SOIL PROPERTIES AND LANDFORMS

757

Tesfahunegn et al. (2011), and those cases classified soil groups with a correct classification rate of 62% in Cambodia by Oberth¨ ur et al. (2000). CONCLUSIONS

Fig. 2 Visualization of discriminant functions in discriminating the group variables as stable, eroding and aggrading sites by plotting the individual scores of each case (sampling point).

minant analysis correctly classified 12 of the 12 crossvalidated cases for the eroding sites, with a correct classification rate of 100%. In the cross validation, only 11 of the 12 cases for the stable sites and 12 of the 14 cases for the aggrading sites were correctly classified with a correct classification rate of 91.7% and 85.7%, respectively. Generally, about 94.7% of the original grouped cases and 92.1% of the cross-validated cases were correctly classified by the discriminant analysis, which proved the acceptability of the field level survey (identification) of different erosion statuses. The overall correct classification rate of the original grouped cases in this study was higher (94.7%) as compared to the cases used to classify soil quality categories (90%) in the northern Ethiopia reported by

Soil properties significantly varied across different erosion statuses in the study catchment. Such variability was described by soil silt, clay, OC, TN, Pav , CEC, Fe, Zn and exchangeable bases, which were higher in the aggrading sites as compared to the stable and eroding sites. It was also generalized that most soil properties significantly varied across the landforms even in the same erosion status. The mountain and central ridge landforms showed the lowest soil nutrients and fine soil particles, whereas the highest values were found in the reservoir followed by the valley landform. Soil properties in the eroding sites were found below their critical levels required for sustainable crop production. Nevertheless, the soil nutrients such as TN, Fe and Zn were below their critical levels across the three erosion-status sites even though their values were lower in the eroding sites, indicating that introduction of appropriate management interventions to restore these soil nutrients across all sites is crucial. The discriminant analysis also identified silt, CEC, BD, Zn, TN, Pav and Fe as the best indicators to monitor variability among the erosion statuses in the study catchment. About 95% of the sites were correctly classified by the discriminant function, indicating that the classification of field-based surveying of the erosion statuses is acceptable and crucial for decision making.

TABLE VIII Classification of cases (sampling points) in group variables (erosion-statuses) by the discriminant analysis Case

Groupa)

Discriminant classification of predicted group membership Eroding

Original grouped

Cross-validatedd)

a) The

Eroding Stable Aggrading Total Eroding Stable Aggrading Total

11b) 0 0 11 12b) 0 0 12

Stable 0 12b) 1 13 0 11b) 2 13

Aggrading

Correct classification rate

1 0 13b) 14 0 1 12b) 13

% 91.7 100 92.9 94.7c) 100 91.7 85.7 92.1e)

weighted cases (sampling points) used in the analysis of the stable, eroding, and aggrading erosion statuses (group variables) were 12, 12, and 14, respectively. b) Number of cases correctly classified by the discriminant analysis. c) Overall 94.7% of original grouped cases were correctly classified. d) In the cross validation, each case is classified by the functions derived from all cases other than that case (sampling point in a given erosion status). e) Overall 92.1% of cross-validated cases were correctly classified.

758

Appropriate interventions should be introduced to improve and maintain soils of the eroding sites across all the landforms using field-based classification of erosion statuses, with special attention to the mountain and central ridge landforms that possessed severely degraded soil properties. ACKNOWLEDGEMENTS The authors gratefully acknowledge the financial support by Deutscher Akademischer Austausch Dienst (DAAD)/Deutsche Gesellschaft f¨ ur Internationale Zusammenarbeit (GIZ) (Germany) through the Centre for Development Research (ZEF), University of Bonn (Germany), and the support of Aksum University (Ethiopia) for the first author’s field work. The authors deeply appreciate the assistance offered by the local farmers and extension agents during the field study. The authors are also grateful to the editor and anonymous reviewers for their comments, suggestions and corrections for improving this paper. REFERENCES Anderson J M, Ingram J S I. 1989. Tropical Soil Biology and Fertility: A Handbook of Methods. CAB International, Wallingford. Arshad M A, Lowery B, Grossman B. 1996. Physical tests for monitoring soil quality. In Doran J W, Jones A J (eds.) Methods for Assessing Soil Quality. SSSA Spec. Publ., Vol. 49. Soil Science Society of America, Madison. pp. 123–142. Baruah T C, Barthakur H P. 1999. A Text Book of Soil Analysis. Vikas Publishing House Pvt. Ltd., New Delhi. Bationo A, Waswa B, Kihara J, Kimetu J (eds.). 2007. Advances in Integrated Soil Fertility Management in Sub Saharan Africa: Challenges and Opportunities. Springer, Dordrecht. Battison L A, Miller M H, Shelton I J. 1987. Soil erosion and corn yield in Ontario. I. Field evaluation. Can J Soil Sci. 67: 731–745. Black C A (ed.). 1965. Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. American Society of Agronomy, Soil Science Society of America, Madison. Blake G R, Hartge K H. 1986. Bulk Density. In Klute A (ed.) Methods of Soil Analysis. Part 1. Physical and Mineralogical Methods. 2nd Ed. Agronomy Monograph 9. American Society of Agronomy, Madison. pp. 363–375. Brady N C, Weil R R (eds.). 2002. The Nature and Properties of Soils. 13th Ed. Prentice-Hall Inc., New Jersey. Bremmer J M, Mulvaney C S. 1982. Total nitrogen. In Page A L, Miller R H, Keeney D R (eds.) Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. Agronomy Monograph 9, American Society of Agronomy, Madison. pp. 595–624. Brierley G J, Fryirs K A, Jain V. 2006. Landscape connectivity: the geographic basis of geomorphic applications. Area. 38: 165–174. Coyne M S, Thompson J A. 2006. Math for Soil Scientists. Thomson Delmar Learning, Clifton Park, New York.

G. B. TESFAHUNEGN et al.

Cruickshank J G, Cruikshank J G. 1972. Soil Geography. David and Charles Pub. Ltd., Newton Abbot. El-Swaify S A, Hurni H. 1996. Transboundary effects of soil erosion and conservation in the Nile basin. Land Husbandry. 1: 6–21. EMWR (Ethiopia Ministry of Water Resources). 2002. Assessment and Monitoring of Erosion and Sedimentation Problems in Ethiopia. Final Report, RODECO Consulting, GmbH. Hydrology Studies Department, Addis Ababa, Ethiopia. Eswaran H, Lal R, Reich P F. 2001. Land degradation: an overview. In Bridges E M, Hannam I D, Oldeman L R, Penning de Vries F W T, Scherr J S, Sombatpanit S (eds.) Response to Land Degradation. Enfield Science Publishers Inc., Enfield. pp. 20–35. Everitt B S, Dunn G. 1992. Applied Multivariate Data Analysis. Oxford University Press, New York. Fageria N K. 2009. The Use of Nutrients in Crop Plants. CRC Press, Boca Raton. FAO (Food and Agriculture Organization of the United Nations). 1986. Ethiopian Highland Reclamation Study. Final report, FAO, Rome. FAO (Food and Agriculture Organization of the United Nations). 1998. The Soil and Terrain Database for Northeastern Africa. FAO, Rome. Dale M R T, Fortin M J. 2005. Spatial Analysis: A Guide for Ecologists. Cambridge University Press, Cambridge. Gee G W, Bauder J W. 1986. Particle-size analysis. In Klute A (ed.) Methods of Soil Analysis. Part 1. 2nd Ed. American Society of Agronomy, Soil Science Society of America, Madison. pp. 383–411. Gregorich E G, Anderson D W. 1985. Effects of cultivation and erosion on soils of four toposequences in the Canadian prairies. Geoderma. 36: 343–354. Haileslassie A, Priess J, Veldkamp E, Teketay D, Lesschen J P. 2005. Assessment of soil nutrient depletion and its spatial variability on smallholders’ mixed farming systems in Ethiopia using partial versus full nutrient balances. Agric Ecosyst Environ. 108: 1–16. Hair J F, Anderson R E, Tatham R L, Black W C. 1992. Multivariate Data Analysis with Readings. Prentice Hall Professional Technical Reference. Macmillan Publishing Company, New York. Haregeweyn N, Poesen J, Deckers J, Nyssen J, Haile M, Govers G, Verstraeten G, Moeyersons J. 2008. Sediment-bound nutrient export from micro-dam catchments in Northern Ethiopia. Land Degrad Dev. 19: 136–152. Heckrath G, Djurhuus J, Quine T A, Van Oost K, Govers G, Zhang Y. 2005. Tillage erosion and its effect on soil properties and crop yield in Denmark. J Environ Qual. 34: 312–324. Hurni H. 1993. Land degradation, famine, and land resource scenarios in Ethiopia. In Pimentel D (ed.) World Soil Erosion and Conservation. Cambridge University Press, Cambridge. pp. 27–62. Jones J B Jr., Eck H V. 1973. Plant analysis as an aid in fertilizing corn and grain sorghum. In Walsh L M, Beaton J D (eds.) Soil Testing and Plant Analysis. Revised Ed. Soil Science Society of America, Madison. pp. 349–364. Kosmas C, Gerontidis S, Marathianou M. 2000. The effect of land use change on soils and vegetation over various lithological formations on Lesvos (Greece). Catena. 40: 51–68. Kreznor W R, Olson K R, Banwart W L, Johnson D L. 1989. Soil, landscape, and erosion relationships in a northwest Illinois watershed. Soil Sci Soc Am J. 53: 1763–1771. Lal R. 1999. Erosion impact on soil quality in the tropics. In Lal R (ed.) Soil Quality and Soil Erosion. CRC Press, Boca

ASSESSING SOIL PROPERTIES AND LANDFORMS

Raton. pp. 285–317. Landon J R. 1991. Booker Tropical Soil Manual: A Handbook for Soil Survey and Agricultural Land Evaluation in the Tropics and Subtropics. Booker Tate, Thame. Legendre P, Dale M R T, Fortin M J, Gurevitch J, Hohn M, Myers D. 2002. The consequences of spatial structure for the design and analysis of ecological field surveys. Ecography. 25: 601–615. Mekonen K, Tesfahunegn G B. 2011. Impact assessment of soil and water conservation measures at Medego watershed in Tigray, northern Ethiopia. Maejo Int J Sci Technol. 5: 312– 330. Muchena F N, Onduru D D, Gachini G N, Jager A. 2005. Turning the tides of soil degradation in Africa: capturing the reality and exploring the opportunities. Land Use Pol. 22: 23–31. Nye P H, Stephens D. 1962. Soil fertility. In Wills J G (ed.) Agriculture and Land Use in Ghana. Oxford University Press, London. pp. 127–143. Oberth¨ ur T, Dobermann A, White P. 2000. The rice soils of Cambodia. II. Statistical discrimination of soil properties by the Cambodian Agronomic Soil Classification System (CASC). Soil Use Manage. 16: 20–26. Olsen S R, Sommers L E. 1982. Phosphorus. In Page A L Miller R H, Keeney D R (eds.) Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. Agronomy Monograph 9, American Society of Agronomy, Madison. pp. 403– 430. Papiernik S K, Lindstrom M J, Schumacher J A, Farenhorst A, Stephans K D, Schumacher T E, Lobb D A. 2005. Variation in soil properties and crop yield across an eroded prairie landscape. J Soil Water Conserv. 60: 388–395. Papiernik S K, Lindstrom M J, Schumacher T E, Schumacher J A, Malo D D, Lobb D A. 2007. Characterization of soil profiles in a landscape affected by long-term tillage. Soil Till Res. 93: 335–345. Papiernik S K, Schumacher T E, Lobb D A, Lindstrom M J, Lieser M L, Eynard A, Schumacher J A. 2009. Soil properties and productivity as affected by topsoil movement within an eroded landform. Soil Till Res. 102: 67–77. Pennock D J 1998. New perspectives on the soil erosion–soil quality relationship. In IAEA (ed.) Use of 137 Cs in the Study of Soil Erosion and Sedimentation. IAEA Publ. IAEATECDOC-1028. IAEA, Vienna. pp. 13–25. Pennock D J, Anderson D W, De Jong E. 1994. Landscapescale changes in indicators of soil quality due to cultivation in Saskatchewan, Canada. Geoderma. 64: 1–19. Pierce F J, Lal R. 1994. Monitoring the impact of soil erosion on crop productivity. In Lal R (ed.) Soil Erosion Research Methods. 2nd Ed. Soil and Water Conservation Society, Ankeny. pp. 235–263. Plaster E J. 1992. Soil Science and Management. 2nd Ed. Delmar Publishers Inc., Albany. Rhoades J D. 1982. Cation exchange capacity. In Page A L, Miller R H, Keeney D R (eds.) Methods of Soil Analysis. Part 2. Agronomy Monograph 9. American Society of Agronomy, Madison. pp. 149–157. S´ anchez P A, Couto W, Buol S W. 1982. The fertility capability soil classification system: interpretation, applicability and modification. Geoderma. 27: 283–309.

759

Saavedra C. 2005. Estimating spatial patterns of soil erosion and deposition in the Andean region using geo-information techniques: A case study in Cochabamba, Bolivia. Ph.D. Dissertation, Wageningen University. Schimel D S, Kittel T G F, Knapp A K, Seastedt T R, Parton W J, Brown V B. 1991. Physiological interactions along resource gradients in a tallgrass prairie. Ecology. 72: 672–684. Shukla M K, Lal R, Ebinger M. 2006. Determining soil quality indicators by factor analysis. Soil Till Res. 87: 194–204. SPSS Inc. 2011. Statistical Package for Social Sciences. Release 18.0. SPSS Inc., Chicago. Stone R, Gilliam J W, Cassel D K, Daniels R B, Nelson L A, Kleiss H J. 1985. Effect of erosion and landscape position on the productivity of Piedmont soils. Soil Sci Soc Am J. 49: 987–991. Stoorvogel J J, Smaling E M A. 1990. Assessment of Soil Nutrient Depletion in Sub-Saharan Africa, 1983–2000. Report 28. DLO Winand Staring Center for Integrated Land, Soil and Water Research, Wageningen. Tamene L. 2005. Reservoir siltation in the drylands of northern Ethiopia: Causes, source areas and management options. Ph.D. Thesis, University of Bonn. Tesfahunegn G B. 2010. Tied Ridging as Insitu Rainwater Harvesting Method for Improving Sorghum (Sorghum bicolor L.) Yield at Abergelle Area, Tigray Regional State, Ethiopia. Lap LAMBERT Academic Publishing GMBH & Co. KG, Saarbrucken. Tesfahunegn G B, Tamene L, Vlek P L G. 2011. A participatory soil quality assessment in Northern Ethiopia’s Mai-Negus catchment. Catena. 86: 1–13. Thomas D J, Cassel D K. 1979. Land-forming Atlantic Coastal Plain soils: Crop yield relationships to soil physical and chemical properties. J Soil Water Conserv. 34: 20–24. Thomas G W. 1982. Exchangeable cations. In Page A L, Miller R H, Keeney D R (eds.) Methods of Soil Analysis. Part 2. Chemical and Microbiological Properties. Agronomy Monograph 9, American Society of Agronomy, Madison. pp. 159– 165. Thomas G W. 1996. Soil pH and soil acidity. In Sparks D L (ed.) Methods of Soil Analysis. Part 3. Chemical Methods. Soil Science Society of America, Madison. pp. 475–490. Ulrich A, 1947. Plant analysis a diagnostic procedure. Soil Sci. 55: 101–112. UNEP and UNESCO (United Nations Environment Program and United Nations Educational, Scientific and Cultural Organization). 1980. Provision Map of Present Land Degradation Rate and Present State of Soil. Food and Agriculture Organization of the United Nations (FAO), Rome. Wan Y, E1-Swaify S A. 1997. Flow-induced transport and enrichment of erosional sediment from a well-aggregated and uniformly-textured Oxisol. Geoderma. 75: 251–265. Withers P J A, Lord E I. 2002. Agricultural nutrient inputs to rivers and groundwaters in the UK: Policy, environmental management and research needs. Sci Total Environ. 282283: 9–24. Zougmor´ e R, Mando A, Stroosnijder L. 2010. Benefits of integrated soil fertility and water management in semi-arid West Africa: an example study in Burkina Faso. Nutr Cycl Agroecosys. 88: 17–27.