Contents lists available atScienceDirect
Catena
journal homepage:www.elsevier.com/locate/catena
Impact of landslides on soil characteristics: Implications for estimating their age
E. Van Eynde
a,⁎, S. Dondeyne
a, M. Isabirye
b, J. Deckers
a, J. Poesen
aaDepartment of Earth and Environmental Sciences, KU Leuven, Celestijnenlaan 200E, Leuven 3001, Belgium
bDepartment of Natural Resources Economics, , Busitema University, Tororo, Uganda
A R T I C L E I N F O
Keywords:
Chronosequence Topsoil Soil fertility Soil organic carbon Rock fragments
A B S T R A C T
The slopes of Mount Elgon, a complex volcano at the border between Uganda and Kenya, are frequently affected by landslides with disastrous effects on the livelihood of its population. Since local people greatly depend on the land for crop production, this paper examines if and how fast physico-chemical characteristics in landslide scars recover. A chronosequence of 18 landslides covering a period of 103 years was sampled in order to explore differences between topsoil inside and outside landslide scars. For each landslide, two topsoil samples were taken within the landslide and two in nearby undisturbed soils to compare their physico-chemical character- istics. Samples inside the landslides were located at the transition zone between the depletion and accumulation zone, which is situated at the contact line between the plan concave and plan convex section of the landslide. No differences were found for available phosphorus, Ca2+, Mg2+content or for thefine earth texture. Recent landslides had however lower content of soil organic carbon(OC) and K+, and higher content of rock fragments and Na+than the adjacent soils. Soil OC content increased significantly with age and reached levels of the corresponding undisturbed soils after ca. 60 years. Older landslides had even higher OC contents than soils adjacent to the landslide. Hence landslide scars act as local carbon sink. We suggest that the occurrence of rock fragments in the topsoil is a useful indicator for mapping past landslides. Moreover, the difference in soil OC content between landslide scars and adjacent soil could be used for estimating the age of landslides in data-poor regions.
1. Introduction
Landslides affect humans in many ways leading to socio-economic disarray, casualties and environmental damage. Landslides induce significant soil loss, sediment deposition and result in the mixing of soil material. Previous studies showed that topsoil characteristics are altered in the process of soil movement. Most authors report lower contents of nitrogen and soil organic carbon of soils located inside landslides compared to nearby undisturbed soils (e.g. Dalling and Tanner, 1995; Guariguata, 1990; Manjusha, 1990; Reddy and Singh, 1993; Zarin and Johnson, 1995a). In terms of soil texture,Zarin and Johnson (1995a)found lower clay content in landslides in a montane forest in Puerto Rico. As far as soil nutrients are concerned, results are less clear. Most authors reported lower content of available phosphorus (P), and exchangeable Ca2+, Mg2+ and K+ (e.g. Guariguata, 1990;
Manjusha, 1990; Reddy and Singh, 1993). However, several researchers have challenged these findings. Shrumpf et al. (2001) argued that landslides can bring deeper, less weathered and therefore more
nutrient-rich material to the surface, which leads to a possible improvement of soil fertility. Similarly,Adams and Sidle (1987), found higher pH and higher exchangeable Ca2+concentrations in three recent landslides in Southeast Alaska. AlsoManjusha (1990)observed a higher pH in landslides in Kumaun Himalaya despite lower nutrient concen- trations. Recently,Cheng et al. (2016)observed higher rock fragment content and pH, but lower concentrations of organic carbon due to landslide deposition in Central Taiwan.
Temporal trends in characteristics of soils located in landslide scars have been reported for organic carbon (OC), nitrogen (N), available P, exchangeable basic cations and soil texture, with an increase in soil fertility over time (Lundgren, 1978; Manjusha, 1990; Reddy and Singh, 1993; Zarin and Johnson, 1995a, 1995b). However, soil fertility restoration typically takes a few decades, as reported for the Uluguru mountains in Tanzania byLundgren (1978), who found lower OC and clay content inside landslides compared to undisturbed soils 7 years after the occurrence of the landslide. Similarly, Zarin and Johnson (1995a, 1995b)found no full restoration of the OC and Ca2+, Mg2+and
http://dx.doi.org/10.1016/j.catena.2017.05.003
Received 10 December 2016; Received in revised form 6 April 2017; Accepted 5 May 2017
⁎Corresponding author.
E-mail addresses:elisevaneynde@gmail.com(E. Van Eynde),stefaan_dondeyne@yahoo.co.uk(S. Dondeyne),isabiryemoseswb@gmail.com(M. Isabirye), seppe.deckers@kuleuven.be(J. Deckers),jean.poesen@kuleuven.be(J. Poesen).
Available online 24 May 2017
0341-8162/ © 2017 Elsevier B.V. All rights reserved.
MARK
K+ contents in the topsoil over a period exceeding 55 years. Other studies did notfind a significant spatial or temporal trend in values of soil properties. The large spatial heterogeneity of parent material within a landslide may be an important obstacle for capturing significant trends (Walker and Shiels, 2013).
Most research reporting on the impact of landslides on soil fertility characteristics has been conducted in Central America. In contrast, whereas the East African highlands are prone to landslides and have been the object of several studies (e.g.Broothaerts et al., 2012; Davies, 1996; Jacobs et al., 2016; Kitutu et al., 2009; Knapen et al., 2006;
Mugagga et al., 2012; Ngecu and Mathu, 1999; Van Den Eeckhaut et al., 2009), few studies have analysed their impact on soil fertility char- acteristics (Lundgren, 1978).
As on Mount Elgon landslides frequently occur in cropland the objective of this study was to assess their impact on the physico- chemical characteristics of the topsoil. In particular, we attempted to determine how long it would take for soil characteristics inside land- slides to recover from the disturbance by landsliding to the level of nearby undisturbed soils. Based on previous studies, we focussed on the effect of landsliding on the content and recovery rate of topsoil organic carbon (OC), exchangeable Ca2+, Mg2+, K+, Na+, soil texture and rock fragments.
2. Materials and methods
2.1. Study area
The study area is in Bududa district (0°5′45″- 1°7′22″N, 34°16′18″ – 34°32′67″E; Uganda), one of the eight districts covering the south-
western part of Mount Elgon (Fig. 1). Based on a reconnaissancefield survey in March 2014, Bududa district was found to have the largest number of landslides from all eight districts in the Mount Elgon region and was therefore chosen for this study. Mount Elgon (4321 m a.s.l.) is an extinct complex volcano at the border between Uganda and Kenya and was mainly built up during the Pliocene (Davies, 1952). The Elgon succession consists of the Basement complex, followed by Pre-Elgon volcanic activity with alkaline intrusions, the building of a shield volcano (Mount Elgon) followed by erosion and deposition processes (Davies, 1956). Mt. Elgon is a central volcano with its summit at 4321 m (Ollier and Pain, 2000). Based on SRTM data (Shuttle Radar Topography Mission (SRTM) voidfilled 1 arc-second global elevation data (~ 30 m) available athttps://earthexplorer.usgs.gov/; accessed 14 March 2017), we determined that from north to south the volcanic cone is ca. 70 km across and 50 km from west to east. In Uganda its base is at ca. 1200 m while in east Kenya it is at ca. 1800 m a.s.l. It is an alkaline stratovolcano built of a succession of agglomerate deposits, principally nephelinite and olivine basalt lava separated by ash layers (Ollier and Pain, 2000; Westerhof et al., 2014). The volcanic cone has a radial drainage but despite being of Miocene age, instead of having planezes theflanks consist essentially of a series of steps or structural terraces (Ollier and Pain, 2000). Bududa district is dominated by granites from the Pre-Cambrium complex together with an alkaline intrusion from Pre-Elgon volcanic activity (GTK Consortium, 2012). Only some small parts at the northern and eastern side of this district are covered by volcanic materials. The harmonized soil map of Africa (Dewitte et al., 2013) indicates that the dominant Reference Soil Groups are Nitisols, Acrisols, Ferralsols and Luvisols.
Annual rainfall ranges from 1500 to 2200 mm with a seasonally
Fig. 1.Location of the dated and sampled landslides in Bududa district on Mt. Elgon, Uganda (topography based on SRTM data downloaded from USGShttps://earthexplorer.usgs.gov/).
E. Van Eynde et al. Catena 157 (2017) 173–179
bimodal distribution. The two peaks of the rainy season fall in March–June and October–November. Average air temperature varies little over the year and is 23 °C at Mbale (1140 m a.s.l., 1°04′38″N 34°10′52″E) (National Environment Management Authority Uganda, 2010).
In 2014, Bududa had a population density of around 850 inhabi- tants per km2. From 2002 to 2014, average annual population growth rate was 3.6% (UBOS, 2014).
The highest parts of Mount Elgon (i.e. above ca. 2000 m a.s.l.) are covered by montane rainforest and afro-alpine meadows and are part of a National Park established in 1992. In the lower parts, land-use is dominated by mixed farming systems with both perennial crops (coffee, banana, some fruit trees and eucalyptus) and annual crops (maize, cassava, beans, groundnuts and potatoes) (Oyana et al., 2014).
2.2. Selection of landslides for soil sampling
During a reconnaissancefield survey in March 2014 during which many interviews and observations were made, 303 landslides were inventoried in Bududa district. From this inventory, 18 landslides were selected based on their initiation year (as reported by local people), type, dimensions and land use.
The landslides were chosen such that the widest possible period could be covered–which turn out to be 103 years with the oldest from 1911 and the youngest from 2014 – (Fig. 1) and with at least one landslide per decade. Furthermore, only deep-seated (i.e. shear plane at a depth > 3 m) rotational landslides with a clear depletion and deposition zone were selected. As such, the selected landslides are expected to have the largest impact on soil characteristics. The selected landslides were typically ca. 50 m wide and 100 m long, with the affected area exceeding 5000 m2and are representative for landslides
with grave impact on people's livelihood. All these landslide scars were under cropland and the slope inclination ranged from 13° to 37°, with an average value of 25°.
2.3. Soil sampling and analysis
Between July and October 2014, soil samples of the upper 10 cm were taken along a line in the transition zone between the depletion (erosion) and the deposition zone of the landslide, and parallel to the overall contour line (Fig. 2). The transition zone could be easily identified in thefield by observing the plan concave slope section of the depletion zone and the plan convex slope section of the deposition zone. Two composite soil samples were taken inside the landslide and two outside in nearby cropland, at 5 to 10 m away from the edge of the landslide scar. Each composite sample consisted offive bulked sub- samples: one taken at the centre of the sampling plot and four at 5 m around it.
The dried soil samples were analysed at the Soil Service of Belgium.
Soil organic carbon content (OC) was determined by the Walkley and Black method (ISO, 1998). Available P and exchangeable basic cations (Ca2+, Mg2+, K+, Na+) content were determined after extraction with ammonium lactate (0.1 N ammonium lactate and 0.4 N acetic acid;
pH 3.75). pH was measured in 1 M KCl (ISO, 2005). Percentage rock fragment (R) was determined by weighing the fragments with a diameter between 3 and 20 mm separated with sieves. Particle size distribution of the fine fraction (clay, silt and sand content) was measured following the hydrometer method (Gee and Bauder, 1986).
2.4. Data analysis
All statistical analyses were done using the statistical software R (R Fig. 2.Examples of three landslides having a different age on Mount Elgon, Bududa district, Uganda. (a) Downslope view of a major landslide that occurred on June 2012 (photo taken 25 Oct. 2012, JP). The black arrows show positions inside and outside the landslide scar (in the transition area between the landslide depletion and deposition zone) where topsoil samples were taken; (b) Uphill view of the same landslide two years later (photo Sep. 2014, JD); (c) Deposition zone of a landslide from 1970 (photo Sep. 2014, JP); (d) Transition area between depletion and deposition zone of a landslide that occurred in 1911 (photo Sep. 2015, JP). Note large rock fragments at the soil surface in the landslide zone (photo c and).
development core team, 2008). Through pairwise comparison, the hypothesis that physico-chemical characteristics from within and out- side the landslides would be equal was tested with the Mann-WhitneyU test (Dytham, 2011). Subsequently, this hypothesis was also tested for three age classes, whereby the age classes of the landslides were defined as: young (1997–2014), middle-aged (1962–1982) and old (1911–1952) yielding six landslides per age class.
Changes of the soil characteristics over time were analysed using orthogonal regression. Orthogonal regression was used, rather than the ordinary least square method, for being a more robust estimator of linear relations than the least square method and as the regression line can be interpreted in both directions (i.e. Y = f(X) just as well as X = f (Y)). Moreover, in orthogonal regression both the response and predictor variables can contain measurement errors, which in our case is appropriate as it cannot be assumed that the estimated ages of the landslides would have no error (Golub and Van Loan, 1980; Carroll and Ruppert, 1996; Petras and Podlubny, 2010).
The recovery rate of the physico-chemical characteristics was checked byfirst plotting the absolute values of the soil characteristics of the sample taken from within the landslides against the age of the landslide. Secondly, the relative changes were checked by plotting the difference between the average of the soil characteristic of the samples taken outside minus inside the landslide.
3. Results
3.1. Impact of landslides on topsoil characteristics
The average values of the physico-chemical soil characteristics from inside and outside each landslide are presented inTable 1. The soil pH of all samples are all moderately acidic and there is no significant difference between the landslide and the adjacentfields.
Fine earth texture ranged from sandy loam to clay following the FAO textural classes (FAO, 2006). Comparing soil particle-size distribu- tion from inside with outside the landslide does not show systematic differences, neither when all soil samples are grouped nor when landslides are divided into three age classes. This observation is confirmed by a Mann-Whitney test for texture indicating no significant differences between disturbed and undisturbed soils for the three landslide age classes.
When making the overall comparison between the physico-chemical soil characteristics from inside with outside the landslide, only sig- nificant differences were found for soil OC, exchangeable Na+and rock fragment content (R). When grouped per age class, significant differ- ences were found for soil OC, K+, Na+and R (Fig. 3.).
Overall the soil OC content in the topsoil is lower inside the landslide scars than outside. This difference is however only significant for the youngest landslide group (Fig. 3a). The K+content within young landslide scars is also lower than outside the landslide (Fig. 3b). In contrast, Na+ content was significantly higher inside the landslides than outside; the difference being most pronounced in the youngest landslides (Fig. 3c). Rock fragment content was overall also higher Table 1
Mean values of physico-chemical topsoil (0–10 cm) characteristics inside (IN) and outside (OUT) landslides. The data for each landslide is the average of two composite samples consisting of 5 bulked sub-samples. The year of the reportedfirst landslide occurrence is indicated as well.
Year Sample pH OC (%) P (mg/100 g) K (mg/100 g) Mg (mg/100 g) Ca (mg/100 g) Na (mg/100 g) Rocks (%) Sand (%) Clay (%) Textural
Position Class⁎
1911 IN 5.3 1.8 1 12 17 137 < 0.9 5.6 45.7 29.8 SCL
OUT 6.5 1.1 2 101 30 98 < 0.9 7.7 46.4 30.5 SCL
1920 IN 4.9 1.7 2 33 34 97 1.8 9.7 22.4 58.5 C
OUT 4.7 1.5 1 22 20 9 < 0.9 6.0 12.4 73.5 C
1930 IN 5.0 2.3 2 27 44 160 1.2 7.0 21.4 60.5 C
OUT 5.0 1.6 2 45 35 121 < 0.9 1.4 18.4 66.5 C
1944 IN 5.3 1.1 2 13 23 100 < 0.9 35.0 52.4 24.5 SCL
OUT 4.7 0.9 3 9 12 72 < 0.9 25.2 40.4 34.5 CL
1951 IN 5.3 1.2 15 18 22 185 1.5 14.7 55.4 25.5 SCL
OUT 5.1 2.0 6 21 28 171 < 0.9 9.0 45.4 32.5 SCL
1952 IN 5.0 1.7 1 15 20 68 1 5.9 24.4 54.5 C
OUT 5.0 1.6 1 28 19 76 < 0.9 6.0 12.4 67.5 C
1962 IN 5.2 0.9 2 33 30 84 < 0.9 31.2 36.4 42.5 C
OUT 4.9 1.0 3 14 17 115 < 0.9 11.4 25.4 50.5 C
1966 IN 4.9 0.9 2 11 35 127 < 0.9 22.8 38.4 34.5 CL
OUT 5.3 1.2 3 16 19 119 < 0.9 4.4 31.4 35.5 CL
1970 IN 4.9 0.8 2 5 29 119 4.9 28.8 69.4 15.5 SL
OUT 4.6 1.1 1 8 25 100 1.1 17.2 58.4 25.5 SCL
1975 IN 4.9 1.2 1 16 56 250 < 0.9 11.6 38.4 33.5 CL
OUT 5.0 1.3 1 19 44 236 < 0.9 8.2 29.4 41.5 C
1978 IN 4.8 1.0 4 24 31 122 0.9 16.8 17.4 59.5 C
OUT 5.0 1.4 3 13 35 151 < 0.9 7.3 25.4 47.5 C
1982 IN 5.2 1.5 3 24 28 151 1.4 2.6 18.4 61.5 C
OUT 5.4 1.6 4 41 40 131 < 0.9 3.2 14.4 67.5 C
1997 IN 5.2 1.1 5 24 64 342 1.2 71.5 26.4 45.5 C
OUT 5.1 1.7 3 25 40 195 < 0.9 11.7 20.4 48.5 C
1997 IN 4.9 0.8 6 19 54 431 2.6 22.9 25.4 46.5 C
OUT 5.6 1.8 14 48 59 294 1.6 5.0 20.4 51.5 C
2007 IN 5.4 1.6 2 9 20 174 < 0.9 25.1 66.4 19.5 SL
OUT 6.2 1.3 22 45 26 203 < 0.9 18.3 55.4 23.5 SCL
2010 IN 6.4 0.9 40 18 25 552 10.4 22.3 52.4 29.5 SCL
OUT 5.2 2.0 16 18 33 364 1.2 4.7 38.4 36.5 CL
2012 IN 4.9 0.7 6 30 30 135 1.1 14.6 17.4 59.5 C
OUT 6.2 1.4 4 80 40 157 < 0.9 53.58 11.4 71.5 C
2014 IN 5.2 0.8 8 23 53 387 2.8 22.1 16.4 58.5 C
OUT 6.1 1.8 29 92 57 389 1.4 6.4 15.4 59.5 C
⁎FAO textural classes (FAO, 2006) with C = clay, CL = clay loam, SCL = sandy clay loam, SCL = sandy clay loam, SL = sandy loam.
E. Van Eynde et al. Catena 157 (2017) 173–179
inside landslide scars than outside, but the difference seems to decrease with landslide age (Fig. 3d).
3.2. Temporal trends in topsoil characteristics
Significant linear trends over time were only found for soil OC content, both when expressed in absolute terms (OC%) as well as in relative terms (ΔOC%) as shown inFig. 4. The absolute soil OC content increases with age of the landslide (Fig. 4a), whereas the difference in soil OC content between outside and inside the scar decreases (Fig. 4b).
Afterca60 years no difference in OC content could be observed. The five sampling points from landslide scars older than 60 years indicate that soil OC content of the topsoil inside landslides even exceeds the corresponding value of the control samples.
4. Discussion
Our results show that landslides have a significant impact on the soil physico-chemical characteristics of the topsoil, as we found significant differences in OC, K+, Na+and rock fragment content. Young land- Fig. 3.Comparison of box plots for physico-chemical characteristics of topsoils sampled inside (IN) and outside (OUT) landslides.P-values refer to the Mann-WhitneyUtest. R = rock fragment content.
slides were depleted in OC and K+, whereas such differences were not found for the older landslides. On the other hand, young landslides have higher content of rock fragments compared to undisturbed soils.
Soil OC content inside landslides seems to increase with time. The absence of significant differences between disturbed topsoils and adjacent undisturbed topsoils for K+ and rock fragment content for the oldest age class suggests that there may also be a recovery for K+ and rock fragment content. The concentration of K+may increase over time due to the input of crop residues, organic manure, the application of fertilizers, the weathering of primary minerals and the deposition of colluvial material. The decrease of rock content over time may also be caused by the deposition offiner colluvium.
For all landslide age classes the topsoils in landslide scars had higher contents of Na+. In the volcanic deposits of Mount Elgon, the presence of for instance Natrolite (Na2Al2Si3O10.2H2O) has been reported (Gottardi and Galli, 1985) which may explain the higher concentrations of Na+ in recent landslides scars as landslides can cause the re- surfacing of this parent material.
Fine earth texture ranged from sandy loam soils to pure clay. As the study area covers a transition zone between Precambrian complex and volcanic materials, the lithology is heterogeneous explaining the wide range of clay and sand contents. No differences in soil texture were found between landslide and undisturbed topsoils, contrary to what has been reported from studies in montane forest in Puerto Rico (Zarin and Johnson, 1995a). A possible explanation for the absence of differences in texture between landslide topsoils and adjacent topsoils may be that the soil profiles in the study area are deeply weathered with diffuse horizon boundaries (e.g. Knapen et al., 2006). As a result, soil disturbance by landslides may not be deep relative to the soil profile, hence no different texture will be observed in disturbed topsoils compared to adjacent undisturbed topsoils.
We have shown that an increase of rock fragment content (R) in topsoils within landslides is a consequence of landslides, which was also shown byCheng et al. (2016). In the specific landscape around Mount Elgon, these relatively old tropical soils are highly weathered, as they underwent physical and chemical disintegration during a long period and as they consist of veryfine particles (Van Wambeke, 1992). The presence of rock fragments is therefore unexpected and can be used as an indicator in thefield for detecting soil disturbances by landslides.
For the absence of spatial differences in divalent cations Ca2+and Mg2+between landslide-affected and undisturbed areas one possible explanation may be that these topsoil nutrients are more attached to the soil particles because of their charges than the monovalent ions K+and Na+. The same goes for available phosphorus for which the sesqui- oxides of these tropical soils have a high affinity. These results would suggest that landslides have a smaller impact on divalent cations and phosphorus, as these nutrients are better adsorbed, hence less differ- ences can be detected among disturbed and undisturbed soil since measured concentrations are that low.
Fig. 4shows that the soil OC content of topsoils in young landslides is low, i.e. ca. 0.7%, and that the difference with undisturbed soils can be as high as 1.5%. This difference is comparable with the results reported byLundgren (1978)for recent landslides in cultivated areas of Tanzania.Fig. 4b also suggests that it takes ca. 60 years under current crop management practices for the topsoil in landslide-affected zones to recover from the depletion of organic carbon. Likewise, the regression line infig. 4b indicates that after ca. 60 years the OC content in the topsoil inside landslides exceeds that of the undisturbed area. Landslide scars are typically plan-concave features in the landscape that concen- trate and store surface runoff. Hence, these scars also collect sediments, which can be expected to be significant since soil erosion rates can be high on these cultivated slopes in the humid tropics (e.g.Greenland and Szabolcs, 1994; Greenland et al., 1997; Lal, 1999; Stoorvogel and Smaling, 1998). The concentration offine earth material may explain the observed decrease of rock fragment content over time. Moreover, the colluvial deposits can be a direct source of additional OC or a source of nutrients, which in turn leads to a higher biomass production within the landslide and therefore potentially increases the amount of organic matter returned to the soil. Finally, these concavities concentrate and collect both surface and subsurface runoff which may also create conditions in which organic matter decay is slower than in the undisturbed areas adjacent to the landslides.
The significant positive trend of OC content in landslide-affected zones with the age of the landslide also suggests that the OC content can be used as a simple indicator for the age of a landslide. The latter holds for soil samples taken in a thoroughly disturbed zone of the landslide, such as found in most transition zones between the depletion and the accumulation area. This may be very useful to determine the frequency of landslide occurrences in regions with very few data on the timing of landslides. Such landslide frequency analysis is crucial for producing landslide hazard maps.
Fig. 4.Relation between landslide age (1 to 103 years) and (a) absolute OC content and (b) changes in organic carbon (ΔOC) content in the topsoil of landslide scars. The trend lines obtained by Orthogonal Linear Regression are given together with their equation and root mean square error (RMSE). TheΔOC is defined as the difference between OC content outside and inside the landslide scar. PositiveΔOC values indicate higher OC content outside the scar, negativeΔOC values indicate higher OC content inside the scar.
E. Van Eynde et al. Catena 157 (2017) 173–179
5. Conclusions
The impact of landslides on topsoil characteristics of cropland was examined by comparing topsoil properties within and outside landslide scars and by investigating temporal trends of these properties in landslide scars. Landslides were found to induce a decrease in soil organic carbon (OC) and an increase of rock fragment (R) content, and both an increase or a decrease of the contents of other nutrients. OC content in topsoils recovers after landslide initiation. Moreover, old landslides (exceeding 60 years) even reach higher OC contents than the adjacent undisturbed soils.
The nutrient retention capacities of tropical soils with low cation exchange capacity is expected to be inherently low, which is amplified by the occurrence of landslides since the OC content was found to be smaller than that of undisturbed soils. For this reason the decrease in OC content in the topsoil will lead to a potential decrease in crop yield for farmers' plots located in recent landslide scars, since organic matter is an important source for the retention of plant nutrients. Moreover, farmers who apply inorganic fertilizers in recent landslides, will probably not have the same benefits as the farmers applying these fertilizers on undisturbed soils outside landslide scars since OC content is higher in the latter. These findings suggest that an important management strategy could be to increase OC content in landslide- affected areas, either by adding organic matter (e.g. manure) or by increasing the organic matter input resulting from vegetation growth in the landslide scar.
Landslides also induce an increase in rock fragment content (R) of the topsoil which may pose a challenge for agricultural management (e.g. tillage operations).
Our study indicates that topsoil properties within and outside landslide scars can be used as indicators to asses landslide age and hence also landslide frequency and hazard. A larger R content can be used as afield indicator for a recent landslide occurrence in tropical weathered soils, whereas the OC content can be used as an indicator for the age of landslides.
Acknowledgments
This research was conducted in the framework of a collaborative research project between the KU Leuven and Busitema University as part of a TEAM project called “Sustainable land use and resilient livelihoods in the landslide-prone region of Mount Elgon, Uganda” funded by VLIRUOS (TEAM Uganda 2013 ZEIN2013PR398). This actual study was alsofinanced by a travel grant by VLIRUOS. A special word of thanks toBahati JorumandBetty Namazzi, for their assistance during thefield work.
References
Adams, Paul, Sidle, Roy C., 1987. Soil conditions in three recent landslides in Southeast Alaska. For. Ecol. Manag. 18, 93–102.
Broothaerts, N., Kissi, E., Poesen, J., Van Rompaey, A., Getahun, K., Van Ranst, E., Diels, J., 2012. Spatial patterns, causes and consequences of landslides in the Gilgel Gibe catchment. SW. Ethiopia. Catena 97, 127–136.
Carroll, R.J., Ruppert, D., 1996. The use and misuse of orthogonal regression in linear errors-in-variables models. Am. Stat. 50 (1), 1–6.
Cheng, C.H., Hsiao, S.C., Huang, Y.S., Hung, C.Y., Pai, C.W., Chen, C.P., Menyailo, O.V., 2016. Landslide-induced changes of soil psyciochemical properties in Xitou, Central Taiwan. Geoderma 265, 187–195.
Dalling, J.W., Tanner, E.V.J., 1995. An experimental study of regeneration on landslides in montane rain forest in Jamaica. J. Ecol. 83, 55–64.
Davies, K.A., 1952. The Building of Mount Elgon (East Africa). The Government of Uganda, Entebbe. (62 pp.
Davies, K.A., 1956. The Geology of Part of South-East Uganda. The Government of Uganda, Entebbe (76 pp.).
Davies, T.C., 1996. Landslide research in Kenia. J. Afr. Earth Sci. 23, 541–545.
Dewitte, O., Jones, A., Spaargaren, O., Breuning-Madsen, H., Brossard, M., Dampha, A.,
Deckers, J.A., Gallali, T., Hallet, S.H., Jones, R.J.A., Kilasara, M., Le Roux, P., Micheli, E., Montanarella, L., Thombiano, L., van Ranst, E., Yemefack, M., Zougmore, R., 2013. Harmonisation of the soil map of Africa at the continental scale. Geoderma 211, 138–153.
Dytham, C., 2011. Choosing and Using Statistics: A biologist's Guide. John Wiley & Sons.
FAO, 2006. Guidelines for Soil Description, fourth ed. Food and Agriculture Organization of the United Nations, Rome.
Gee, G.W., Bauder, J.W., 1986. Particle-size analysis. p. 383–411. In: Klute, A. (Ed.), Methods of soil analysis. Part 1, 2nd. Agron. Monogr. 9. ASA and SSSA, Madison, WI.
Golub, G.H., Van Loan, C.F., 1980. An analysis of the total least squares problem. SIAM J.
Numer. Anal. 17 (6), 883–893.
Gottardi, G., Galli, E., 1985. Natural Zeolites. Springer, Berlin Heidelberg, pp. 52.
Greenland, D.J., Szabolcs, I., 1994. Soil Resilience and Sustainable Land Use. CAB International, Wallingford, UK, pp. 576.
Greenland, D.J., Gregory, P.J., Nye, P.H., 1997. Land Resources: On the Edge of the Malthusian Precipice? CAB International, New York, pp. 151–161.
GTK Consortium, 2012. Explanation of the Geology of Sheets NA-36-11 and NA-36-12 (Mbale and Kapenguria) 1:250 000, Uganda. Department of Geological Survey and Mines (DGSM), Entebbe (109 pp).
Guariguata, M.R., 1990. Landslide disturbance and Forest regeneration in the upper Luquillo Mountains of Puerto Rico. J. Ecol. 78, 814–832.
ISO, 1998. ISO 14235:1998. Soil quality—Determination of organic carbon by sulfochromic oxidation. International Standards Organization.
ISO, 2005. Soil quality–Determination of pH. International Standards Organization ISO 10390:2005.
Jacobs, L., Dewitte, O., Poesen, J., Delvaux, D., Thiery, W., Kervyn, M., 2016. The Rwenzori Mountains, a landslide-prone region? Landslides 13, 519–536.
Kitutu, M.G., Muwanga, A., Poesen, J., Deckers, J., 2009. Influence of soil properties on landslide occurrences in Bududa district, eastern Uganda. Afr. J. Agric. Res. 4 (7), 611–620.
Knapen, A., Kitutu, M.G., Poesen, J., Breugelmans, W., Deckers, J., Muwanga, A., 2006.
Landslides in a densely populated county at the footslopes of Mount Elgon (Uganda):
characteristics and causal factors. Geomorphology 73, 149–165.
Lal, R., 1999. Erosion Impact on Soil Quality in the Tropics. Soil Quality and Soil Erosion, Chapter 16. CRC Press, London, pp. 283–317.
Lundgren, L., 1978. Studies of soil and vegetation development on fresh landslide scars in the Mgeta Valley, Western Uluguru Mountains, Tanzania. Geografiska Annaler. Series A, Phys. Geogr. 60, 91–127.
Manjusha, J., 1990. A study on soil and vegetation changes after landslide in Kumaun Himalaya. Proc. Indian natn. Sci. Acad. 4, 351–360.
Mugagga, F., Kakembo, V., Buyinza, M., 2012. A characterization of the physical properties of soil and the implications for landslide occurrence on the slopes of Mount Elgon, Eastern Uganda. Nat. Hazards 60, 1113–1131.
National Environment Management Authority Uganda, 2010. State of the environment.
The Government of Uganda, Kampala. (online,http://www.nemaug.org/reports/
national_state_report_2010.pdf, Accessed 10 August, 2015).
Ngecu, W.M., Mathu, E.M., 1999. The El-Nino-triggered landslides and their socioeconomic impact on Kenya. Environ. Geol. 38, 277–284.
Ollier, C., Pain, C., 2000. The Origin of Mountains. Routledge, London & New York.
Oyana, T.J., Kayendeke, E., Bamutaze, Y., Kisanga, D., 2014. Afield assessment of land use systems and soil properties at varied landscape positions in a fragile ecosystem of Mount Elgon, Uganda. African Geographical Review 34, 83–103.
Petras, I., Podlubny, I., 2010. Least squares or least circles? A comparison of classical regression and orthogonal regression. Chance 23 (2), 38–42.
Reddy, V.S., Singh, J.S., 1993. Changes in vegetation and soil during succession following landslide disturbance in the Central Himalaya. J. Environ. Manag. 39, 235–250.
Shrumpf, M., Guggenberger, G., Valarezo, C., Zech, W., 2001. Tropical montane rain forest soils. Development and nutrient status along an altitudinal gradient in the South Ecuadorian Andes. Erde 132, 43–59.
Stoorvogel, J.J., Smaling, E.M.A., 1998. Research on soil fertility decline in tropical environments: integration of spatial scales. Nutr. Cycl. Agroecosyst. 50, 151–158.
UBOS, 2014. National population and housing census 2014–Provisional results. Ugandan Bureau of Statistics, Kampala. (online:http://www.ubos.org/onlinefiles/uploads/
ubos/NPHC/NPHC%202014%20PROVISIONAL%20RESULTS%20REPORT.pdf, Accessed 11 August, 2015).
Van Den Eeckhaut, M., Moeyersons, J., Nyssen, J., Abraha, A., Poesen, J., Haile, Mitiku, Deckers, J., 2009. Spatial patterns of old, deep-seated landslides: a case-study in the northern Ethiopian highlands. Geomorphology 105, 2239–2252.
Van Wambeke, A., 1992. Soils of the Tropics: Properties and Appraisal. CTA Wageningenpp. 335.
Walker, L.R., Shiels, A.B., 2013. Physical Causes and Consequences. Landslide Ecology.
Cambridge University Press, Cambridge, pp. 46–81.
Westerhof, A.B., Härmä, P., Isabirye, E., Katto, E., Koistinen, T., Kuosmanen, E., Lehto, T., Lehtonen, M.I., Mäkitie, H., Manninen, T., Mänttäri, I., Pekkala, Y., Pokki, J., Saalmann, K., Virransalo, P., 2014. Geology and geodynamic development of Uganda with explanation of the 1: 1,000,000 scale geological map. Geological Survey of Finland, Special Paper 55. Espoo.
Zarin, Daniel J., Johnson, Arthur H., 1995a. Base saturation, nutrient cation, and organic matter increases during early pedogenesis on landslide scars in the Luquillo Experimental Forest, Puerto Rico. Geoderma 65, 317–330.
Zarin, Daniel J., Johnson, Arthur H., 1995b. Nutrient accumulation during Primary Succession in a Montane Tropical Forest, Puerto Rico. SSSAJ 59, 1444–1452.
