Geoderma 265 (2016) 187–195

Contents lists available at ScienceDirect

Geoderma

journal homepage: www.elsevier.com/locate/geoderma

Landslide-induced changes of soil physicochemical properties in Xitou,

Central Taiwan
Chih-Hsin Cheng a,⁎, Sheng-Che Hsiao a, Yu-Sheng Huang a, Chih-Yu Hung a, Chuang-Wen Pai b,
Chiou-Pin Chen b, Oleg V. Menyailo c
a
School of Forestry and Resource Conservation, National Taiwan University, Taipei 106, Taiwan
b
NTU Experimental Forest, National Taiwan University, Nantou 513, Taiwan
c
Institute of Forest SR RAS, Krasnoyarsk 660036, Russia

a r t i c l e i n f o a b s t r a c t

Article history: Steep mountain terrain, fractured geological environments, and intense precipitation events are primary factors
Received 14 September 2015 that contribute to frequent landslides in mountainous regions. Landslides exert an immense impact on forest eco-
Received in revised form 17 November 2015 systems and substantially change soil properties; understanding these changes is crucial to facilitating subse-
Accepted 22 November 2015
quent forest revegetation and management. In this study, the effects of landslide on soil physicochemical
Available online 4 December 2015
properties were investigated in a moist montane forest ecosystem in Xitou, Central Taiwan. We established a
Keywords:
dataset comprised historical soil survey data obtained in 1976 and data from soil samples taken in 2012 at the
Soil morphology same locations to compare differences in soil properties after landslide deposition, and also conducted soil sam-
Pedology pling along a landslide/nonlandslide affected sequence to determine how the degree of landslide deposition af-
Soil organic carbon fected soil physicochemical properties. The results indicated that rock fragment content, soil pH value, bulk
Total nitrogen density, inorganic carbon, and base saturation increased following landslide deposition and that severe landslide
Exchangeable cations deposition caused more substantial increases. By contrast, the thicknesses of the O and A horizons, soil organic
carbon, total nitrogen, and cation-exchange capacity significantly decreased following landslides; these de-
creases were more substantial with increasing degree of landslide deposition. Exchangeable potassium, calcium,
magnesium, and available phosphorus content, however, were unaffected by landslide deposition. The results
demonstrated that soil physicochemical properties were significantly altered after landslide deposition; these re-
sultant changes, particularly in regard to high soil pH value, poor structure, and low soil organic carbon and total
nitrogen, are expected to influence functions in forest ecosystems.
© 2015 Elsevier B.V. All rights reserved.

1. Introduction
Landslides are critical natural disturbances in tropical and temperate
areas that shape mountainous areas and redistribute sediment to less
steep areas. The outward and downward gravitational movement of
earth material can be immense and significantly alters forest ecosystems,
including soil properties, light and moisture regimes, and plant structures
and communities; landslides can also cause catastrophic damage to
infrastructure and property as well as human casualties (Sidle and
Ochiai, 2006; Restrepo et al., 2009). Although most landslides
are naturally occurring phenomena triggered by heavy rains or
earthquakes, rural development-related activities such as the devel-
opment of roads and trails, agricultural cultivation, and land-clearing
on steep terrain also have the potential to influence landslide
⁎ Corresponding author at: R207 Forestry Hall, School of Forestry and Resource
Conservation, National Taiwan University, Taipei 106, Taiwan.
E-mail address: chengch@ntu.edu.tw (C.-H. Cheng).
activities (Sidle et al., 2006; Sassa and Canuti, 2009). Landslides are
currently recognized as disasters rather than as natural processes
(Sidle and Ochiai, 2006).
Regarding changes in soil properties after landslides, decreases in
soil organic carbon (SOC) and available nutrients are common because
parent material at deeper layers that generally contains less SOC and
available nutrients is exposed (the landslide scar) and redistributed to
the surface (the landslide deposition area) (Adams and Sidle, 1987;
Guariguata, 1990; Singh et al., 2001; Wilcke et al., 2003; Sparling et al.,
2003). However, the movement of parent material via landslides can
also replenish nutrients in acidic and nutrient-poor soils that have lost
nutrients because of weathering (Schrumpf et al., 2001). Therefore,
changes in soil properties following landslides might be site-specific
and depend on the type of landslide; namely, on the movement of
parent material and the thickness of the failure plane or landslide deposi-
tion. Understanding the effects that landslides have on soil properties has
become increasingly critical; such information is essential for subsequent
forest revegetation and management (Sidle et al., 2006; Sassa and Canuti,
2009) and may enable foresters to understand how changes in soil

http://dx.doi.org/10.1016/j.geoderma.2015.11.028
0016-7061/© 2015 Elsevier B.V. All rights reserved.
188
properties affect vegetation growth, plant communities, and net primary
production in landslide areas (Walker et al., 1996; Shiel et al., 2008;
Restrepo et al., 2009; Dislich and Huth, 2012).
Because of its steep mountain terrain, fractured geological environ-
ments, earthquakes, and heavy precipitation from typhoon events,
Taiwan is one of the areas most prone to landslides in the world
(Chang and Slaymaker, 2002; Lin et al., 2003; Tsai et al., 2010). For ex-
ample, the Chi-Chi earthquake in 1999 caused landslides measuring
11,300 ha, and the subsequent Typhoon Toraji in 2001 additionally
caused landslides exceeding 22,300 ha in Central Taiwan (Chang and
Slaymaker, 2002).Precipitation falling measured 2200 mm over a
three-day period during Typhoon Morakot in 2009 also caused
45,000 ha of landslides in Southern Taiwan (Tsai et al., 2010). These
damages from landslides have arisen mitigating strategies on land
utilization and soil and water conservation to prevent further dete-
rioration (Lu et al., 2001; Tsai et al., 2010). Surprisingly, little is
known about the effects of landslides on soil properties in Taiwan;
such effects clearly require further evaluation. In addition, many
landslides in Taiwan have involved sedimentary rock; however,
previous studies have mainly been conducted in metamorphic and
igneous regions (Guariguata, 1990; Wilcke et al., 2003; Reddy and
Singh, 1993).
This study evaluated the effects of landslides on soil physicochemical
properties in a moist montane forest ecosystem in Xitou, Central Taiwan
Hundreds of landslides occurred after Typhoon Troaji in 2001. Most
landslides were debris-flow type and caused significant downslope
depositions in the basin areas. In this study, we investigated how soil
properties changed after landslides and how the degree of landslide
deposition affected these properties. We established a dataset that
compared differences in soil properties between a 1977 survey and a
current soil survey of soil samples obtained in areas affected and unaf-
fected by landslides and conducted sampling at a series of sites with
varying degrees of landslide deposition to examine how landslide
deposition affected soil physicochemical properties. A variety of soil
physiochemical properties were examined in this study to enhance un-
derstanding regarding changes in soil properties caused by landslides
and to provide practical information for forest revegetation, develop-
ment, and management.
Fig. 1. Locations of study site in Xitou, central Taiwan. Black circles represent the historical soil survey sites. Red stars represent the landslide/nonlandslide sequence sites. Shadows indicate the
landslide areas identified by aerial photos after Typhoon Toraji in 2001.
C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
2. Materials and methods
2.1. Study site
The study was conducted in Xitou, Central Taiwan (23°40′ N,
120°53′ E) (Fig. 1). The site is located in the basin of a valley surrounded
by steep mountains on three sides (mean slope angles 40°, Chang et al.,
2010). Lingtou Mountain is on the southern side of The Valley. The
Phoenix Mountain Range is on the eastern side and stretches northward,
and the Neishupi Mountain Range is on the western side and stretches
northward. The elevation from The Valley basin to the mountain crest
ranges between 900 and 2000 m. The bedrock is the Guizhulin and
Nanchuang formations, which were formed in a shallow marine environ-
ment in the late Miocene and composed of mixed layers of sandstone and
shale. Foraminifera and shellfish fossils are abundant in the rocks and
have lent them calcareous properties (Ho, 1988). The overlying soils are
developed in the sandstone and shale colluvium and have a sequence of
horizons designated O–A–B–C (Chen and Chiang, 1995).
The mean annual temperature over the last 30 years has been 16.8 °C
(at 1100 m asl) and the mean annual precipitation has been 2635 mm.
The site was classified as subtropical wet forest or montane mesic to wet
forest (Holdridge, 1967). The native vegetation was originally dominated
by Machilus and Castanopsis trees (Gu et al., 2005) but was replaced by
forest plantings over the past 100 years, particularly in the basin of The
Valley, where the slope is gentler (Cheng et al., 2013a). The major tree
species that have been planted are Cryptomeria japonica (L. f) D. Don
(Japanese cedar), Chamaecyparis formosensis Matsum. (Taiwan red
cypress), Taiwania cryptomerioides Hayata, and Cunninghamia lanceolata
Hayata. The native broadleaf forests have been kept only on steep slopes
for watershed protection.
High-intensity precipitation from typhoons, steep mountain terrain,
fragile geologic formation and earthquakes render the area to be highly
prone to landslide events (Huang, 2006; Chang et al., 2010). In 2001,
Typhoon Toraji triggered hundreds of landslides measuring 182 ha
and covering 7.3% of the total area in Xitou (2480 ha in total area;
Huang, 2006) (Fig. 1). Most landslides were debris-flow type and caused
by shallow translational movements, characterized by the liquification of
the soil and rock fragment. Major economic and social damage was
 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195 189

Fig. 2. The design of sampling sites along a landslide/nonlandslide affected sequence in Xitou, central Taiwan. Site 1 and Site 2 were the nonlandslide affected soils. Site 3, Site 4, and Site 5 were
the landslide affected soils, from shallow landslide deposition at Site 3 to severe landslide deposition at Site 4 and Site 5. Site3, Site 4, and Site 5 were all disturbed by landslide deposition after
Typhoon Toraji in 2001, and an additional small-scale landslide deposits were also reported at Site 4 after Typhoon Morakot in 2009.

sustained in basin areas where the infrastructure and forest plantations
were devastated by landslide (debris flow) deposition. We therefore
confined our study to areas affected by landslide deposition; landslide
scars and channels on steep slopes were not included.
2.2. Sampling design
Two data sets were used in this study. The first dataset comprised
historical soil survey data obtained in 1976 (Ho, 1977) and data from
soil samples obtained in 2012 at the same locations; the resampled
soil samples were taken both from areas that were affected and unaffect-
ed by landslide deposition. These data were analyzed to compare changes
in soil properties after landslide deposition. The second dataset was used
to evaluate changes in soil physicochemical properties along a landslide/
nonlandslide affected sequence to determine how the degree of landslide
deposition affected the soil physicochemical properties (Figs. 1 and 2).
We included soil survey data carried out by Ho (1977) in the first
sampling set; Ho conducted a soil survey featuring 16 soil profiles in
Xitou in 1976. By analyzing archival aerial photographs dating to
1975, a new soil survey was conducted in 2012 to obtain soil samples
that were affected and unaffected by landslides at the same locations.
Aided by original field notes, site drawings, and photographs, the
resampled soil samples were obtained from within 15 m of the original
soil pits. Three soil samples affected by landslide deposition after
Typhoon Troraji in 2001 and one sample that was not affected by land-
slide deposition were analyzed. The three landslide-deposited soil
samples were obtained from a Ginkgo biloba plantation, a Japanese
cedar plantation in the No. 158 compartment, and a Japanese cedar plan-
tation in the No. 60–3 compartment. The nonlandslide affected soil sam-
ple was obtained from a red cypress plantation. Although all three
landslide affected soils were affected by landslide deposition, the soil in
the No. 60–3 compartment showed the most severe changes, as its vege-
tation was destroyed by landslide deposition. Compared to one sample by
Ho (1977), soil samplings in 2012 were conducted in triplicate to account
for possible spatial variation, and collected at depths of 0–10 cm and 10–
20 cm to enable direct comparison with the survey data obtained by Ho
(1977). According to the results of an experiment conducted by Ho
(1977), the stand ages for the three landslide affected soil samples were
56, 35, and 6 years for soil samples from the G. biloba plantation, the
No. 158 compartment, and the No. 60–3 compartment, respectively,
compared with 92, 75, and 40 years in 2012. Although soil properties
may be affected by stand development, changes in soil properties caused
by stand development were assumed to be minor compared with the
effects of landslide deposition, which exerts a tremendous and long-
lasting impact (Zarin and Johnson, 1995; Singh et al., 2001). In addition,
Cheng et al. (2013a) reported no clear age-related pattern in soil carbon
storage in Japanese cedar plantations in Xitou. Shutou and Nakane
(2004) also found that changes in mineral soil properties with stand
development became slower and leveled off after 20 years in Japanese
cedar plantations in Japan.
For the second sampling set, soil samples were obtained from a total
of five sites, from Site 1 (S1) to Site 5 (S5), along a landslide/nonlandslide

Table 1
Changes in soil physicochemical properties between 1977 and 2012 soil survey data.

Sitea Yeara A horizon Soil color Rock fragments pH SOC (g kg−1) Total N (g kg−1) CECpot (cmol (+) kg−1)
(thickness, cm) (moist, 0–10 cm) (%, vol.)
0–10 cm 10–20 cm 0–10 cm 10–20 cm 0–10 cm 10–20 cm 0–10 cm 10–20 cm

Nonlandslide affected soil

Red cypress 1976 18 10YR2/2 30 5.1 4.5 89.4 26.8 8.0 2.0 36.1 21.9
2012 13 10YR2/2 25 4.8 3.8 67.3 43.2 6.7 4.3 29.2 24.2

Landslide affected soils

No. 158 1976 20 10YR2/2 15 4.9 4.7 61.0 30.0 6.0 3.0 21.5 18.4
2012 3 10YR3/3 25 6.0 5.8 25.6 25.4 2.3 2.6 12.0 12.4
Ginkgo 1976 8 10YR2.5/1 12 5.5 4.7 36.5 16.8 3.0 2.0 31.8 21.8
2012 2 2.5Y3/3 30 6.0 5.8 17.6 15.7 1.9 1.8 13.0 12.6
No. 60–3 1976 13 10YR3/3 4 4.3 4.0 75.4 26.7 8.0 3.0 30.5 20.4
2012 2 10YR4/3 35 5.2 4.9 20.2 11.8 2.3 1.4 13.7 10.7
a
Site and year are referred to Ho (1977).
190 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195

Table 2
Soil morphological and physical characteristics in the nonlandslide affected (S1 and S2) and landslide affected (S3, S4, and S5) zones in Xitou, central Taiwan.

Vegetation type O horizon A horizon Soil texture (0–10 cm) Soil color (0–10 cm) Rock fragmenta MWHCa
(thickness, cm) (thickness, cm) Dry moist (%, vol)
Sand (%) Silt (%) Clay (%) (%)

S1 Cryptomeria japonica 1.3 10.8 60 32 8 2.5Y3/2 10YR3/1 10 73

S2 Cryptomeria japonica 1 10.1 60 33 7 2.5Y3/1 10YR2/1 15 68
S3 Cryptomeria japonica b1 3.3 63 34 3 2.5Y5/4 10YR3/3 15 49
S4 Grass and shrubs b1 b1 67 30 3 2.5Y6/2 2.5Y4/4 50 23
S5 Grass and shrubs b1 3.1 61 36 3 2.5Y4/2 2.5Y3/2 40 34
a
Field estimated rock fragment content and MWHC (maximum water holding capacity) for 0–20 cm soils.

affected sequence (Fig. 2). S1 was located in a 90-year old Japanese cedar
plantation stand (No. 30 compartment) and was slightly away from the
toe of the slope. Archival aerial photographs (dating to 1975) and the
compartmental record showed that S1 had not been disturbed by
landslide events in recent years. Both S2 and S3 were located in a
61-year-old Japanese plantation stand (No. 173 compartment) and
were closer to the toe of the slope compared with the S1 site. S2 and S3
were approximately 100 m apart and did not differ visually in plantation
stand characteristics. However, the compartmental record and on-site
field observations indicated that S3 was disturbed by shallow landslide
deposition after Typhoon Toraji in 2001. Thus, the top 0–10 cm soil at
S3 was overlaid by small-scale pile up from landslide deposition, whereas
no landslide deposition was exhibited at the S2 site. S4 and S5 were locat-
ed in landslide deposition fans. The sites used to be the Japanese cedar and
cypress forest plantations, respectively, but were overlaid by thick
landslide deposition when the areas weres hit by Typhoon Toraji. Aerial
photographs and on-site field observations indicated substantial stand-
replacing disturbances following landslide. At S4 site, an additional

Fig. 3. Soil physicochemical properties along a landslide/non-landslide sequence in Xitou, central Taiwan. Site 1 and Site 2 were the nonlandslide affected soils. Site 3, Site 4, and Site 5 were
the landslide affected soils. (a) pH, (b) bulk density, (c) soil organic carbon (SOC), (d) inorganic carbon (IC), (e) total nitrogen (N), (f) potential cation exchange capacity (CECpot), (g) base
saturation (BS) percentage, (h) exchangeable Ca, (i) exchangeable K, (j) exchangeable Mg, and (k) available P. Data show means ± standard deviation (n = 3). Different capital letters
mean significant difference at p b 0.05 for 0–10 cm surface soil and small letter for 10–20 cm subsurface soil.
 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
Fig. 3 (continued).
small-scale landslide deposits were also reported after Typhoon Morakot
in 2009. The effects of landslide deposition were still visible at both S4 and
S5, as vegetation at these sites mainly comprised annual and perennial
plants and no large woody plants were present.
We established three 20 × 20 m sampling plots at each designated
site. In each sampling plot, mineral soil samples were collected and
pooled from three locations at fixed depths at the surface 0–10 cm
and subsurface 10–20 cm. The thickness of the O (mainly in Oi and
Oe) and A horizon were determined in the field based on hand-dug
pits in each sampling plot. The rock fragment content based on volume
and soil color were determined in the field based on the Munsell Soil
Color charts. A stainless steel core with a known internal volume was
used to measure soil bulk density after the soil samples were oven-
dried at 105 °C. The maximum water holding capacity (MWHC) was
measured by taking soil cores, soaking them with water for two days,
draining them, and then weighing them.
2.3. Soil analyses
The collected mineral soil samples were air-dried and ground to pass
through a 2-mm sieve. The soil samples were analyzed according to the
method indicated by Carter and Gregorich (2007). Soil pH value was
measured at a 1:2.5 soil:solution ratio (in deionized water) by using a
pH electrode after shaking for 1 h. Soil texture was measured using
the Bouyoucos hydrometer method. Soil total carbon and total nitrogen
191
were determined using an elemental analyzer (Perkin Elmer 2200,
Massachusetts, USA) after the soil samples were further ground using
a ball grinder (Oscillating Mill MM400 by Retsch, Pennsylvania, USA).
SOC was determined after removing the inorganic carbon by using
0.1 N HCl. Soil inorganic carbon was calculated as the differential
between the soil total carbon and SOC (Wang et al., 2012). Exchange-
able Ca, Mg, and K and the potential cation exchange capacity (CECpot)
were determined after leaching the soil with 1 M ammonium acetate
at pH 7.0 and subsequently replaced by 2 M KCl. Exchangeable Ca, Mg,
and K (extract of 1 M NH4OAC) were determined using an atomic
absorption spectrophotometer (Sensa AA, GBC, Victoria, Australia),
and the NH+ 4 in the KCl extracts was determined using the distillation
and titration method. The base saturation percentage was calculated as
the ratio of exchangeable Ca, Mg, and K to CECpot. Available phosphorus
was determined by extracting soil with the Mehlich-3 solution;
phosphate concentration was measured using the molybdenum blue
method (Mehlich, 1984).
In the measurements of Ho (1977), only SOC, total nitrogen, CEC,
and pH values were reported. Thus, we compared these values at
both dates. Ho (1977) measured soil organic nitrogen and total
nitrogen by using the Walkley–Black and Kjeldahl methods, respec-
tively, whereas we used an elemental analyzer. We assumed that
differences between soil properties on the two dates were due to
landsliding effects rather than differences in measuring methods
(Wang et al., 2012).
192 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
2.4. Statistical analysis
The normality and homogeneity of the original analytical data was
first examined using the Shapiro–Wilk test and the Levene test, respec-
tively. Data that were not normally distributed or lack of homogeneity
were converted logarithmically before performing a one-way analysis of
variance (ANOVA). The one-way ANOVA of the experimental results
was conducted using R 2.14.1 and the Tukey HSD post hoc test to
determine whether soil properties from samples affected and unaffected
by landslide deposition differed significantly (p b 0.05).
3. Results
3.1. Changes in soil properties between 1977 and 2012
Comparing soil properties at the same locations between 1977 and
2012, changes in soil properties were apparent in the landslide deposi-
tion zones (Table 1). Following landslide deposition, the soils exhibited
weak profile development, characterized by a thinner A horizon thick-
ness and lighter soil color. The thickness of the A horizon decreased
from 8 to 20 cm in the 1977 soil survey to less than 3 cm in the current
soil survey. Similarly, chemical properties were also markedly altered
after landslide deposition. Landslide deposition resulted in a pH value
raised by 1, from approximately pH 4.9 to approximately pH 5.7, for
both 0–10 cm and 10–20 cm soil samples. The values of SOC, total
nitrogen, and CECpot were reduced by 30% to 70%; soils at the surface
(0–10 cm) exhibited the most substantial reductions. The mean
values of SOC, total nitrogen, and CEC pot at the 0–10 cm surface
were 57.6 g kg−1, 5.7 g kg−1, and 27.9 cmol kg−1 in the 1977 survey
and 20.8 g kg−1, 2.2 g kg−1, and 12.9 cmol kg−1 in the current survey,
respectively.
The soil sample unaffected by landslide deposition exhibited fewer
differences from the original 1977 survey data than did samples in the
landslide deposition zones. A slightly lower pH value and higher SOC,
total N, and CECpot contents were even observed in the 10–20-cm
subsurface soil in 2012 (Table 1).
3.2. Soil properties along a landslide/nonlandslide affected sequence
Differences in soil physicochemical properties along the landslide/
nonlandslide affected sequence further corroborated the observation
that soil properties were substantially altered by landslide deposition
(Table 2; Fig. 3). Soil samples from zones without landslide deposition
exhibited the least rock fragment content; rock fragment content
increased with the degree of landslide deposition, from 10 to 15% at
S1 and S2 to 35–40% at S4 and S5. The thicknesses of the O and A
horizons decreased or were lost in landslide deposition zones, from
1 cm and N10 cm, respectively, at S1 and S2 to being absent at S4 and
S5. The color of the soil samples obtained at S1 and S2 was darker
than was that of soil samples obtained at S4 and S5. Furthermore, soil
samples obtained at S1 and S2 exhibited higher MWHCs. However,
regarding texture class, the soil samples at all sites were relatively
comparable and classified as sandy loam, with the exception of the
soil sample obtained from S4, which showed a slightly higher sand
content.
Soil pH value, bulk density, inorganic carbon, and base saturation
were lowest in zones unaffected by landslide deposition and tended to
increase with the degree of landslide deposition (p b 0.05, Fig. 3). Soil
pH value increased from 5.2 in soil samples unaffected landslide deposi-
tion (S1 and S2) to 8.3 in soils obtained from the landslide deposition
zones (S4 and S5); soil bulk density increased from 0.7 to 1.8 g cm−3;
base saturation increased from 19% to over 100%; and inorganic carbon
increased from negligible to 0.6 and 9.3 g kg−1. By contrast, SOC, total
nitrogen, and CECpot content showed a significantly downward trend
as the degree of landslide deposition increased (p b 0.05, Fig. 3). SOC
decreased from 43 g kg− 1 at S1 and S2 to 3.8 g kg−1 at S4 and S5;
total nitrogen decreased from 24 to 0.7 g kg−1; and CECpot decreased
from 24 to 7.4 cmol kg−1.
Surface soil samples from depths of 0–10 cm generally exhibited
lower soil bulk densities and pH values and higher SOC, total N, and
CECpot than did subsurface soil samples from depths of 10–20 cm.
However, soil samples at S3 where the surface soil was superimposed
by small scale landslide were less consistent. The 10–20 cm subsurface
soil at S3 showed less soil bulk density and higher organic carbon
content, total nitrogen content, and CECpot than did the 0–10 cm surface
soil (Fig. 3).
No significant difference was observed regarding exchangeable Ca,
K, and Mg along the landslide/nonlandslide affected sequence. The
only exception was the exchangeable Ca in the 10–20 cm subsurface
soil, in which the exchangeable Ca content was significantly lower at
S1 compared with that at S5. The available phosphorus ranged between
8.4 and 34.5 mg kg−1. In addition to being comparatively lower at S4 in
the 0–10 cm surface soil, available phosphorus content did not exhibit a
consistent trend along the landslide/nonlandslide affected sequence
and no significant difference was observed regarding the 10–20 cm
subsurface soil samples.
4. Discussion
4.1. Effects of landslide deposition on soil properties in Xitou
By comparing historical and current soil survey samples and soil
samples along a landslide/nonlandslide affected sequence, our results
demonstrated that soil properties were significantly altered after land-
slide deposition. The soil samples in the landslide deposition zones
tended to be higher in rock fragment content, exhibit thinner O and A
horizon thicknesses, and be lighter in color compared with soil samples
unaffected by landslide deposition; furthermore, the soil samples in the
landslide deposition zones also yielded higher soil pH values, inorganic
carbon content, and base saturation values. By contrast, soil samples in
zones without significant landslide deposition corresponded well with
previous soil survey (Chen and Chiang, 1995), and contained less rock
fragment content, had thicker O and A horizon thicknesses, were darker
in color, had lower pH values, and had higher organic carbon, total nitro-
gen, and CEC content compared to the soil samples in the landslide de-
position zone. The properties of soil affected and unaffected by landslide
deposition can thus be summarized in a conceptual model emphasizing
pedogenic (nonlandslide and internal force) and geomorphological
(landslide deposition and external force) processes (Fig. 4). The pedo-
genic process facilitates the accumulation of more organic matter in
surface soil and transforms soils into differentiated and organized soil
profiles, whereas the geomorphological process of deposition overlay
disrupts pedogenic development and leaves behind immature soil
with little development or low profile differentiation (Johnson et al.,
1987; Huggett, 1998; Saint-Laurent et al., 2014).
The changes in soil morphological features caused by landslide
deposition observed in this study were prominent and consistent with
previous studies that have indicated that landslides destroy original
soil profiles (Adams and Sidle, 1987; Dalling and Tanner, 1995; Zarin
and Johnson, 1995; Singh et al., 2001; Wilcke et al., 2003; Sparling
et al., 2003). Soil after landslide deposition shows a higher soil bulk
density and lower MWHC, which can be ascribed to the destruction of
the soil structure and porosity and the admixture of rock fragments
during the process of landslide deposition. Soil in nonlandslide areas,
by contrast, contains less rock fragments and has a loose and porous
structure, which promotes water-holding capacity, aeration, and
drainage. Although the recovery of vegetation after landslides has
been assumed to promote soil development and lower soil bulk
density over time (Guariguata, 1990; Singh et al., 2001; Sparling
et al., 2003), this was not the case in the current study because the re-
covery time was short and the vegetation was not yet fully grown to de-
velop the O and A horizons and lower the soil bulk density.
 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
Fig 4. Conceptual model for soil properties in Xitou, central Taiwan. (a) Landslide deposition occurs during the typhoon and rainstorm, and the geomorphological process disrupts
pedogenic development and leaves behind immature soil with little development or low profile differentiation. (b) The pedological process on the nonlandslide areas allow soils to
develop differentiated and organized profiles and accumulate more organic matter.
Other distinct changes in soil properties following landslides include
decreases in SOC and total nitrogen content (Adams and Sidle, 1987;
Guariguata, 1990; Singh et al., 2001; Wilcke et al., 2003; Sparling et al.,
2003). Such decreases can also be explained by the removal of surface
soil horizons and the deposition of soil and parent material slurry in
the surface soil. In this study, the SOC and total nitrogen content after
landslides was only 20–63% that of the same soil prior to landslides or
even less than 5% in comparing landslide affected soil samples with
soil samples obtained from adjacent locations. Similar dramatic
decreases have also been reported by others (Guariguata, 1990; Singh
et al., 2001; Sparling et al., 2003). For example, Singh et al. (2001)
reported that decreases in SOC and total nitrogen content caused by
landslides in a forest in the Nepali foothills ranged from 31.5 to
7.6 g kg−1 for SOC and from 3.1 to 0.8 g kg−1 for total nitrogen.
Guariguata (1990) contended that landslides reduced SOC and total
nitrogen in subtropical, low-elevation, and moist foothill forests in Puerto
Rico from 40.7 to 1.7 g kg−1 and from 3.7 to 0.5 g kg−1, respectively.
Furthermore, previous research has indicated that the SOC and total
nitrogen content required several decades of development to achieve
the original status before the landslide (Zarin and Johnson, 1995; Singh
et al., 2001). As SOC and total nitrogen content is crucial in soil fertility,
such decreases may serve as a profound influence on vegetation growth,
stand development, biodiversity, and net primary productivity in land-
slide areas (Blaschke et al., 1992; Walker et al., 1996; Shiel et al., 2008;
Dislich and Huth, 2012).
Landslide deposition in Xitou caused the soil to increase in pH value
(Table 2; Fig. 3), primarily because calcareous parent material (inorganic
carbon) was introduced into surface soil (Ho, 1988). However, inorganic
carbon can rapidly weather and was almost nonexistent in the
nonlandslide affected soil. The lower pH values in the nonlandslide
affected soil could also have resulted from the accumulation of soil
193
organic matter to form organic acids (Guariguata, 1990; Jobbágy
and Jackson, 2003). Many studies have also reported higher pH
values after landslides (Adams and Sidle, 1987; Guariguata, 1990;
Singh et al., 2001) but marked increases in soil pH values have not
been reported (Rosser and Ross, 2011). This discrepancy can be ascribed
to the differences in parent material between this and other studies that
are mainly in granitic and igneous rocks.
No significant relationship between the distribution of exchangeable
Ca, Mg, and K and available phosphorus and landslide deposition was
observed in this study. The results differed from previous studies that
have suggested that landslides reduce exchangeable cations and avail-
able phosphorus (Adams and Sidle, 1987; Guariguata, 1990; Zarin and
Johnson, 1995; Singh et al., 2001; Wilcke et al., 2003). These discrepan-
cies can also be attributed to the differences in parent material between
this and other studies. The parent material in Xitou comprises mixed
layers of sandstone and shale formed in a shallow marine environment
and contains abundant amounts of Ca, Mg, K, and P (Ho, 1988). Land-
slides can release exchangeable cations and available phosphorus
from parent material into surface soil and render such soil comparable
with surface soil in nonlandslide areas, where exchangeable cations
and available phosphorus are generally enriched by plant uptake from
deeper layers to surface soil in the form of litterfall (Jobbágy and
Jackson, 2003). Other studies have investigated areas in which the
parent material has contained less Ca, Mg, K, and P than does the parent
material in Xitou; in such areas, landslides do not result in the replenish-
ment of exchangeable cations and available phosphorus in surface soil.
Decreases in CECpot have been observed after landslides (Singh et al.,
2001; Wilcke et al., 2003; Sparling et al., 2003). Given that the samples
share similar soil textures, the lower CECpot after landslides can be
attributed to insufficient SOC to provide binding sites for cations. When
a landslide induces a lower CECpot but the amount of exchangeable
194 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
cations remains similar, the landslide affected soil exhibits a higher BS
(sum of exchangeable Ca, Mg, and K/CECpot), increasing from 40% in
nonlandslide affected soil to over 100% in landslide affected soil. The
overestimated BC (N100%) in the landslide affected soil was ascribed to
the dissolution of calcareous parent material in the extractant.
Furthermore, a BS of 40% is still high compared with other forest soils in
Taiwan (Jien et al., 2009; Cheng et al., 2013b). The high BS value in
Xitou may reflect the inherent calcareous properties of the parent
material.
The results from the landslide/nonlandslide affected sequence sug-
gested that changes in soil properties varied with the degree of land-
slide deposition (Table 2; Fig. 3). Varying in accordance with the depth
from which the parent material covering the ground surface originated,
severe landslide deposition caused more severe alterations. However,
even when deposition is slight, soil properties can vary significantly
(e.g., S2 vs. S3 in this study). In addition, differences in the pile-up process
(e.g., in regard to sandstone vs. shale in this study) may affect soil proper-
ties. The soil samples at S4 and S5, both of which were affected by severe
landslide deposition, exhibited different properties; these differences may
be attributed to the variance in types of parent material. The slightly
higher sand content and lower available phosphorus content at S4 could
have been caused by the introduction of parent material containing a
higher portion of sandstone compared with that at S5, which contained
a higher proportion of shale.
4.2. Environmental implications
Besides being an important factor in affecting soil properties, the
processes of landslide can create extreme spatial heterogeneity in soil
properties in the mountainous regions. Several research, for instance,
has reported that soil pH values ranged from 3.6 to 6.0 in Xitou (Ho,
1977; Chiang et al., 2002; Cheng et al., 2013a). Some of the variability
in fact can be attributed to the influences of landslides that are less
often considered. Hence, when collecting soil samples from mountain-
ous regions, not only the effects of the terrain, vegetation, and parent
materials should be examined but also the landsliding effects need to
be included to have a comprehensive understanding of soil properties
across landscapes.
Previous studies have shown that revegetation failed on the landslides
were attributed to the adverse soil and climatic conditions, such as ex-
treme soil temperature and humidity differences, low soil and nutrient
quantity, unstable soil condition, and low seed bank (Dalling and
Tanner, 1995; Walker and Shiels, 2008). The poor soil structure (e.g.
low MWHC) and low fertility (e.g. high pH and low SOC and N) on the
landsliding areas were also suggested in this study. In particular, the de-
creases in SOC and total N contents can be the major limiting nutrients
in the landslide deposition zones in Xitou, because the concentrations of
exchangeable Ca, Mg, and K and available P were less affected by landslid-
ing. The findings can be informative for local land managers to develop
better managements on the landsliding areas. The low SOC and total N
on the landsliding areas may constrain forest C sequestration which is
one of the key functions for forest ecosystems (Restrepo et al., 2009). If
the frequency and magnitude of landslides were enhanced by human ac-
tivities or by the extreme weather with high-intensity precipitation, the
reductions in forest C sequestration by landslides can thus be a wide-
spread problem in balancing the global atmospheric carbon budget. Fu-
ture studies should continuously monitor the development of soil
properties after landslides and focus on how landslides reduce forest
net primary productivity. A complete carbon cycle assessment between
landslides and the erosion processes should also be conducted (Berhe
et al., 2006; Hilton et al., 2008).
5. Conclusion
Our study demonstrated that the effects of landslide deposition on
soil properties are substantial in Xitou. Landslide processes resulted in
soil that was higher rock fragment content, lower in O and A horizon
thicknesses, and lighter in soil color, as well as higher in soil pH values,
inorganic carbon content, and base saturation. By contrast, in soil in
nonlandslide areas, the effects of pedological development facilitated
the profile development of the O and A horizons; furthermore, soil in
nonlandslide areas also exhibited lower soil pH values and higher
SOC, total N, and CEC pot content. We suggest when collecting soil
samples from mountainous regions, not only the effects of the terrain,
vegetation, and parent materials but also the landsliding effects must
be considered to gain a comprehensive understanding of soil properties
across landscapes. Because landslide processes lead to reductions in
SOC and total nitrogen and poor soil structure, resultant changes may
influence vegetation growth, stand development, biodiversity, and net
primary productivity in landslide areas. If landslides are increased in
frequency and magnitude by human activities or by extreme weather
with high-intensity precipitation, reductions in forest carbon sequestra-
tion by landslides may become a widespread problem in balancing the
global atmospheric carbon budget. Future studies should continually
monitor changes in soil properties after landslides and focus on how
landslides reduce forest net primary productivity.
Acknowledgments
This study was supported by the Ministry of Science and Technology
of Taiwan and a cooperative grant from the National Science Council of
Taiwan and the Siberian Branch of the Russian Academy of Sciences. We
gratefully acknowledge Chang-Ya Chen and Hong-Jai Ruan for their
assistance with the field and lab work.
References
Adams, P.W., Sidle, R.C., 1987. Soil conditions in three recent landslide in Southeast
Alaska. For. Ecol. Manag. 18, 93–102.
Berhe, A.A., Harte, J., Harden, J.W., Torn, M.S., 2006. The significance of the erosion-
induced terrestrial carbon sink. Bioscience 57, 337–346.
Blaschke, P.M., Trustrum, N.A., DeRose, R.C., 1992. Ecosystem processes and sustainable
land use in New Zealand steeplands. Agric. Ecosyst. Environ. 41, 153–178.
Carter, M.R., Gregorich, E.G. (Eds.), 2007. Soil Sampling and Methods of Analysis, Second
Edition Canadian Society of Soil Science, CRC Press and Taylor & Francis Group,
Florida, USA.
Chang, J.C., Slaymaker, O., 2002. Frequency and spatial distribution of landslides in a
mountainous draiage basin: western foothills, Taiwan. Catena 46, 285–307.
Chang, C.S., Wey, T.H., Lai, Y.J., Chen, H.H., 2010. Raifall threshold value of debris flow
disaster in the Sitou region. J. Chin. Soil Water Conserv. 41, 362–371 (in Chinese).
Chen, Z.S., Chiang, H.C., 1995. Soil characteristics, genesis, and classification of some forest
soils in Taiwan. In: Schulte, A., Ruhiyat, D. (Eds.), Proceedings of Soils of Tropical
Forest. Ecosystems 1, pp. 63–78.
Cheng, C.H., Chen, Y.S., Huang, Y.H., Chiou, C.R., Lin, C.C., Menyailo, O.V., 2013b. Effects of
repeated fires on ecosystem C and N stocks along a fire induced forest/grassland
gradient. J. Geophys. Res. 118, 215–225.
Cheng, C.H., Hung, C.Y., Chen, C.P., Pei, C.W., 2013a. Biomass carbon accumulation in aging
Japanese cedar plantations in Xitou, Central Taiwan. Bot. Stud. 54, 60.
Chiang, K.I., Chen, Z.H., Wang, Y.N., Liu, S.W., 2002. Ground covers and soil physical and
chemical properties in seven forest stands of Chitou. J. Exp. For. NTU 16, 37–44 (in
Chinese).
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.
Dislich, C., Huth, A., 2012. Modelling the impact of shallow landslides on forest structure
in tropical montane forests. Ecol. Model. 239, 40–53.
Gu, M.S., Chuang, S.T., Chung, N.J., Chung, S.H., 2005. Vegetation analyses of landslides in
chitou area. Q. J. Chin. Forest 38, 267–278 (in Chinese).
Guariguata, M.R., 1990. Landslide disturbance and forest regeneration in the upper
luquillo mountains of Puerto Rico. J. Ecol. 78, 814–832.
Hilton, R.G., Galy, A., Hovius, N., Chen, M.C., Horng, M.J., Chen, H., 2008. Tropical-cyclone-
driven erosion of the terrestrial biosphere from mountains. Nat. Geosci. 1, 759–763.
Ho, C.P., 1977. The Soil Property Analysis of the National Taiwan University's Experimental
Forest Plantation in Sitou (Thesis for the Institute of Forestry, National Taiwan
University, Taipei, Taiwan (in Chinese)).
Ho, C.S., 1988. An Introduction to the Geology of Taiwan Explanatory Text of the Geologic
map of Taiwan. Central Geological Survey, Taipei, Taiwan.
Holdridge, L.R., 1967. Life zone ecology. Tropical Science Center, Sen Jose, Costa Rica.
Huang, C.Y., 2006. Monitoring the Historical Change of Landslide Areas and Vegetation
Recovery in Chitou Area (Thesis for the Institute of Forestry, National Taiwan University,
Taiepei, Taiwan (in Chinese)).
Huggett, R.J., 1998. Soil chronosequences, soil development, and soil evolution: a critical
review. Catena 32, 155–172.
 C.-H. Cheng et al. / Geoderma 265 (2016) 187–195
Jien, S.H., Chiu, C.Y., Chen, T.H., 2009. Soil characteristics and genesis processes of a
subtropical low-elevation mountain forest in central Taiwan. Taiwan J. For. Sci. 24,
27–40.
Jobbágy, E.G., Jackson, R.B., 2003. Patterns and mechanisms of soil acidification in the
conversion of grasslands to forests. Biogeochemistry 64, 205–229.
Johnson, D.L., Watson-Stegner, D., Johnson, D.N., Schaetzl, R.J., 1987. Proisotropic and
proanisotropic processes of pedoturbation. Soil Sci. 143, 278–292.
Lin, C.W., Shieh, C.L., Yuan, B.D., Shieh, Y.C., Liu, S.H., Lee, S.Y., 2003. Impact of Chi-Chi
earthquake on the occurrence of landslides and debris flows: example from the
Chenyulan River watershed, Nantou, Taiwan. Eng. Geol. 71, 49–61.
Lu, S.Y., Cheng, J.D., Brooks, K.N., 2001. Managing forests for watershed protection in
Taiwan. For. Ecol. Manag. 143, 77–85.
Mehlich, A., 1984. Mehlich 3 soil test extractant: a modification of Mehlich 2 extractant.
Commun. Soil Sci. Plant Anal. 15, 1409–1416.
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.
Restrepo, C., Walker, L.R., Shiels, A.B., Bussmann, R., Claessens, L., Fisch, S., Lozano, P., Negi,
G., Paolini, L., Poveda, G., Ramos-Scharron, C., Richter, M., Valazquez, E., 2009.
Landsliding and its multiscale influence on mountainscapes. Bioscience 59, 685–698.
Rosser, B.J., Ross, C.W., 2011. Recovery of pasture production and soil properties on soil
slip scars in erodible siltstone hill country, Wairarapa, New Zealand. N. Z. J. Agric.
Res. 54, 23–44.
Saint-Laurent, D., Gervais-Beaulac, V., Berthelot, J.S., 2014. Variability of soil properties in
different flood-risk zones and link with hydroclimatic changes (southern Quebec,
Canada). Geoderma 214-215, 80–90.
Sassa, K., Canuti, P. (Eds.), 2009. Landslide—Disaster Risk Reduction. Springer-Verlag, Berlin
Heidelberg.
Schrumpf, 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.
Shiel, A.B., West, C.A., Weiss, L., Klawinski, P.D., Walker, L.R., 2008. Soil factors predict initial
plant colonization on Puerto Rican landslides. Plant Ecol. 195, 165–178.
195
Shutou, K., Nakane, K., 2004. Change in soil carbon cycling for stand development of
Japanese cedar (Cryptomeria japonica) plantations following clear-cutting. Ecol. Res.
19, 233–244.
Sidle, R.C., Ochiai, H., 2006. Landslides: Processes, Prediction and Land use. American
Geophysical Union, Washington, DC, USA.
Sidle, R.C., Ziegler, A.D., Negishi, J.N., Nik, A.R., Siew, R., Turkelboom, F., 2006. Erosion
processes in steep terrain- truths, myths, and uncertainties related to forest management
in Southeast Asia. For. Ecol. Manag. 224, 199–225.
Singh, K.P., Mandal, T.N., Tripathi, S.K., 2001. Patterns of restoration of soil physciochemical
properties and microbial biomass in different landslide sites in the Sal forest ecosystem
of Nepal Himalaya. Ecol. Eng. 17, 385–401.
Sparling, G., Ross, D., Trustrum, N., Arnold, G., West, A., Speir, T., Schipper, L., 2003. Recovery
of topsoil characteristics after landslip erosion in dry hill country of New Zealand, and a
test of the space-for-time hypothesis. Soil Biol. Biochem. 35, 1575–1586.
Tsai, F., Hwang, J.H., Chen, L.C., Lin, T.H., 2010. Post-disaster assessment of landslides in
southern Taiwan after 2009 typhoon Morakot using remote sensing and spatial analysis.
Nat. Hazards Earth Syst. Sci. 10, 2179–2190.
Walker, L.R., Shiels, A.B., 2008. Post-disturbance erosion impacts carbon fluxes and plant
succession on recent tropical landslide. Plant Soil 313, 205–216.
Walker, L.R., Zarin, D.J., Fetcher, N., Myster, R.W., Johnson, A.H., 1996. Ecosystem develop-
ment and plant succession on landslides in the Caribbean. Biotropica 28, 566–576.
Wang, X., Wang, J., Zhang, J., 2012. Comparisons of three methods for organic and
inorganic carbon in calcareous soils of northwestern China. PLoS One 7, e44334.
Wilcke, W., Valladarez, H., Stoyan, R., Yasin, S., Valarezo, C., Zech, W., 2003. Soil properties
on a chronosequence of landslides in montane rainforest, Ecuador. Catena 53, 79–95.
Zarin, D.J., Johnson, A.H., 1995. 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.