Вы находитесь на странице: 1из 9

This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy

Journal of Arid Environments 73 (2009) 306–313

Contents lists available at ScienceDirect

Journal of Arid Environments


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

Changes in vegetation and landscape patterns with altered river water-flow


in arid West China
W. Kong a, d, O.J. Sun b, *, W. Xu a, Y. Chen c
a
State Key Laboratory of Vegetation and Environmental Change, Institute of Botany, Chinese Academy of Sciences, Beijing 100093, China
b
Key Laboratory for Silviculture and Conservation of Ministry of Education, College of Forest Science, Beijing Forestry University, Beijing 100083, China
c
Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi 830011, China
d
Graduate University of Chinese Academy of Sciences, Beijing 100049, China

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

Article history: Landsat TM images of Tarim Basin in western China for 1986, 1999 and 2004 were analyzed for quan-
Received 12 September 2007 tifying the patterns of landscape change relating to changes in water supply. Results showed that
Received in revised form vegetation area and NDVI mostly decreased from 1986 to 1999, and increased from 1999 to 2004, while
2 October 2008
changes in desert area displayed an inverse pattern. Saline alkali soil showed a tendency of increase from
Accepted 3 October 2008
Available online 18 November 2008
1999 to 2004. Spatially, percentage of vegetation area decreased and percentage of desert area increased
with distance from the river in the upper section, while such patterns were not observed for both the
middle and the lower sections. Landscape displayed a pattern of fragmentation from 1986 to 1999 and
Keywords:
Arid regions integration from 1999 to 2004. Shape of vegetation patches tended to become more regular from 1986 to
Groundwater 1999 and more irregular from 1999 to 2004. Our results indicated hydrological control of spatio-temporal
Hydrology variations of vegetation and landscape pattern in arid regions. Water diversion can be effective for raising
Landscape pattern the local groundwater level and improving plant growth, but its effect is largely restricted to areas
TM imagery analysis adjacent to the water pathway.
Tarim River Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction vegetation on groundwater and consequence of declining


groundwater level to vegetation cover and ecosystem functioning
In arid environment, water is the most limiting factor to plant in arid and semiarid regions. For example, using 16-year record of
growth, and the spatio-temporal dynamics of vegetation are plant cover derived from satellite data, Elmore et al. (2006) showed
therefore largely determined by water availability (Li et al., 2001; that the plant community in the Owens Valley of California, USA,
Elmore et al., 2006). Understanding the relationship between water was groundwater dependent and groundwater extraction
supply and spatio-temporal variations of vegetation and landscape adversely affected the plant cover. In the Doñana National Park of
pattern is critically important to developing and implementing Spain, Muñoz-Reinoso (2001) found that declining groundwater
strategies for biodiversity conservation and maintenance of level resulted in changes of vegetation to more xerophytic
ecosystem structure and function in arid regions (Wu and Hobbs, communities over a 30-year period. Changes in riparian vegetation
2002a). of semiarid regions due to declining groundwater level have also
The amount and frequency of rainfall are highly variable and been reported (Stromberg et al., 1996). The above-mentioned
scarce in arid regions, and the direct effects of precipitation on studies have helped with gaining the recognition of the impact of
vegetation growth can be very weak (Chen et al., 2004c; Elmore declining groundwater to vegetation succession and cover change
et al., 2006). Therefore, groundwater plays a predominant role in in arid and semiarid regions. However, little is known on whether
supporting plant communities. In recent years, intensified the process could be reversed by a recovery in groundwater level.
anthropogenic activities have accelerated the depletion of What could be effective mechanisms in restoring groundwater
groundwater worldwide, resulting in rapid decline of vegetation level in arid regions? There appears a need for continued efforts in
cover and changes in plant community structures (Stromberg et al., elucidating the relationship between groundwater level and
1996; Muñoz-Reinoso, 2001; Naumburg et al., 2005; Elmore et al., vegetation in regions of water limitation.
2006). Several recent studies have shown the dependence of In an attempt to conserving declining plant resources and
stabilize the socio-economic development of the arid region, the
Chinese government has in recent years implemented several key
* Corresponding author. initiatives targeted at environmental protection. One of the initia-
E-mail address: sunjianx@bjfu.edu.cn (O.J. Sun). tives is concerned with restoring vegetation in areas of high

0140-1963/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jaridenv.2008.10.003
Author's personal copy

W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313 307

ecological significance through emergency water diversion, such as surrounded by Takelamakan Desert in the south and Kuluktag
that in Tarim Basin western China, where annual precipitation is Desert in the north and is among the most arid areas in China (Chen
often less than 50 mm (Chen et al., 2004c). Because of the scarce et al., 2006). Annual rainfall ranges from 17.4 to 42.0 mm (Chen
rainfall, plant growth in the Tarim Basin is predominantly et al., 2004c).
controlled by groundwater supplies (Chen et al., 2004b,c, 2006; Vegetation is mainly composed of halophytic plants in fami-
Zhuang and Chen, 2006), and much of the local vegetation are lies of Salicaceae, Tamaricaceae, Leguminosae, Apocynaceae and
distributed along the Tarim River or places where water supplies Gramineae (Zhang and Chen, 2004). Tree species is predomi-
are readily available through snowmelts or seasonal flooding. Since nately Populus euphrarica, which forms the most important
1950s, ecological processes along the Tarim River changed drasti- vegetation type along the Tarim River, pure or in mixture with
cally under intensified human activities related to water exploita- Tamarix spp. Shrub species include Nitraria sibirica, Tamarix
tion for socio-economic development of the region (Chen et al., ramosissima, Tamarix hispida, Tamarix spp. and Halimodendron
2003a). Interruptions of water supplies to the lower reaches of the halodendron. Phragmites communis dominates the herbage
Tarim River due to construction of a dam in 1972 at the upper species and co-occurs with other herbaceous plants including
section, the Daxihaizi, greatly affected the vegetation along the Glycyrrhiza inflata and Alhagi sparsifolia, which collectively form
river drainage area, which has been known as a ‘‘Green Corridor’’ the halophytic meadow distributing extensively in this area
separating the Takelamakan and Kuluktag deserts (Chen et al., (Song, 1999). Desert covers the largest area in the lower Tarim
2003a). For the purpose of conserving and restoring the local River, with euphrate poplar forests, Tamarix bush, halophytic
ecosystems, an emergency water diversion project was imple- meadow, desert, saline alkali soil and water body being other
mented for transferring water from the Lake Bosten to the Daxihaizi dominant land cover types.
Reservoir, and then to the downstream section of the Tarim River, A dam was built at Daxihaizi in 1972 for creating a reservoir,
beginning in 2000. In total, 1.8 billion m3 of water was delivered to consequently causing interruptions of water-flow to the down-
the downstream river system of the Daxihaizi Reservoir on six stream river system. In 2000, an emergency water diversion
separate occasions until 2004 (Xu et al., 2007), resulting in effec- project was implemented for transferring water from the Dax-
tively raised groundwater level and apparently improved vegeta- ihaizi Reservoir to the lower reaches. The initial water diversion
tion growth in the river drainage area (Chen et al., 2003b, 2004c). was made through the Qiwenkuor River channel. In 2003,
Field survey along the lower reaches of the Tarim River a second river path, known as the old Tarim River was opened for
following water diversion indicates a strong influence of ground- transferring part of the water in a ‘‘double-bridge’’ water trans-
water on plant community structure and species diversity (Chen ferring scheme for expanding the spatial coverage of the target
et al., 2004c, 2006). However, the effect of hydrology at larger areas (Xu et al., 2007). Until 2004, water diversions were made
spatial scale is not easily obtainable through field investigation. on six separate occasions, delivering a total of 1.8 billion m3
Remote sensing imagery analysis can assist with studying vegeta- water to the downstream river system (Xu et al., 2007), and
tion and landscape patterns at larger spatio-temporal scales raised groundwater level along the river drainage area (Chen
(Palmer and Van Rooyen, 1998; Wolter and White, 2002; Wu et al., et al., 2003b, 2004c).
2002b; Zha et al., 2003; Elliott et al., 2004). There are now many For convenience of data analysis and assessment of spatial
satellite remote sensing-derived vegetation indices available for features of vegetation change during the study periods, we arbi-
characterizing vegetation cover, e.g. the normalized difference trarily divided the whole study area into the upper, middle and
vegetation index (NDVI), the perpendicular vegetation index (PVI), lower sections longitudinally along the river channel. The upper
the enhanced vegetation index (EVI), and the soil adjusted vege- section is from the Daxihaizi Reservoir to Kardayi where the first
tation index (SAVI) (Gilabert et al., 2002). Among those indices, water diversion reached; the middle section is from Kardayi to
NDVI has been the most frequently used for studying changes in Alagan where the old Tarim River and the Qiwenkuor River meet;
vegetation cover in response to anthropogenic and environmental and the lower section extends from Alagan to the terminal lake of
perturbations (e.g. Palmer and Van Rooyen, 1998; Weiss et al., Tarim River (see Chen et al., 2006). Each side of the Qiwenkuor
2004; Anyamba and Tucker, 2005). For coarser spatial resolution River is further divided into different zones based on distance from
(>250 m), NOAA-AVHRR and MODIS offer direct NDVI data sources. the river channel: zone 0, 0–1 km; zone 1, 1–2 km; zone 2, 2–5 km;
For NDVI at finer resolution, Landsat TM or ETM imagery appears to and zone 3, 5–10 km (labeled as E and W, indicating the east and
be widely used products (Zha et al., 2003). west of the river channel).
In this study, we quantified changes in vegetation cover and
other major landscape features in the lower reaches of the Tarim
River from 1986 to 1999 (i.e. during the period of declining water 2.2. Remote sensing products and buffer zone making
supply) and from 1999 to 2004 (i.e. during the period of increasing
water supply) using Landsat TM imagery analysis. The primary Landsat 5 TM images of 1986, 1999 and 2004 were used in this
objectives of our study were to: (a) examine how and in what ways study. All images were multi-spectral data with all bands 1–7 at
regional vegetation and landscape responded to the two contrast- 28.5 m spatial resolution, and acquired between end of July and
ing periods of changes in water supply to the lower reaches of the mid-September (i.e. the period of peak growth of plants in the
Tarim River drainage area; and (b) assess at a relatively larger region). Those images were provided by the China Remote Sensing
spatial scale, the effectiveness of the emergency water diversion Satellite Ground Station (RSGS). Geometric corrections of the
project in restoring the local ecosystem and its management images were made against a geometrically corrected TM image of
implications. 2001 for the same area from the Maryland University (http://glcf.
umiacs.umd.edu/data), which was corrected with ground control
2. Materials and methods points in polynomial transformation method with correction
precision to one pixel. All images were projected in Albers equal
2.1. Study area area conic projection. An ERDAS 8.4 image processing system was
used for all image data processing. Images were selected with
The study area extends from the Daxihaizi Reservoir down- regard to the availability and quality of satellite data, seasonal
stream to the end of Tarim River (39 340 –40 400 N, 87 210 –88 270 E) patterns of vegetation, and occurrence of maximum NDVI to
in central Xinjiang Autonomous Region, western China. This area is ascertain a fair comparison of images obtained at different times.
Author's personal copy

308 W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313

Frequent sandstorms have been a major constraint for selecting The CA was calculated as:
images with close matching of the sampling dates.  
n
X
We drew the Qiwenkuor River way using vector function in 1
CA ¼ aij
ERDAS 8.4 based on the image of 2004. Buffers for the subdivided 10; 000
j¼1
zones relative to the river channel were produced with the buffer
making function in ArcGIS 8.3 and ERDAS 8.4. where aij is the area (m2) of the patch ij.
The PD was calculated as:
2.3. Computation of NDVI
N
PD ¼ ð10; 000Þ  ð100Þ
NDVI is defined by the spectral reflectance in the band 3 (red A
band, 0.63–0.69 mm) and the band 4 (near IR band, 0.76–0.90 mm) where N is the total number of patches in the landscape, and A is
as: the total landscape area (m2).
The shape index was calculated as:
NDVI ¼ ðnear IR band  red bandÞ=ðnear IR band þ red bandÞ
Pij
SHAPE ¼
The original TM images record the radiance value reflected by min pij
ground objects as DN (digital number) value, which is affected
by sensor and the atmosphere (Lillesand and Kiefer, 2000). where Pij is the perimeter of the patch ij in terms of number of cell
Therefore radiometric calibration and atmospheric correction are surfaces and min Pij is the minimum perimeter of the patch ij in
both necessary in NDVI calculation (Zha et al., 2003; Tachiiri, terms of number of cell surfaces.
2005). In this study, the atmospheric correction of NDVI imagery The contagion index was calculated as:

2 " !# " !#3


Pm Pm Pm Pm
i¼1 k¼1 ðPi Þ gik = k¼1 gik  lnðPi Þ gik = k¼1 gik
6 7
6 7
CONTAG ¼ 61 þ 7  100
4 2lnðmÞ 5

was attempted without causing apparent impact, and the where Pi is proportion of the landscape occupied by the patch type
radiometric calibration alone was proven effective in detecting I, and gik the number of adjacencies between pixels of the patch
the effect of altered water-flow on NDVI. The pixel-based NDVI types i and k based on double-count method, and m the number of
was calculated and used for computation of mean NDVI for each patch types present in the landscape.
of the subdivided zones. In an attempt to separate effects of groundwater change and
changes in climate, we analyzed the inter-annual variations of
temperature and precipitation for the study period by using
2.4. Image classification and calculation of landscape indices
meteorological data from the Tikanlik Meteorological Station
(40 380 N, 87420 E) near our study area.
Based on knowledge of cover types, we constructed training
data sets. The study area was classified into six land cover
types, i.e. desert, saline alkali soil, euphrate poplar forest, 3. Results
Tamarix bush, halophytic meadow, and river way, using
supervised classification (Lillesand and Kiefer, 2000). In the 3.1. Classification map
classification training we used the 5–4–3 false color composite
of the TM images which matches most closely with the nature Classification maps of 1986, 1999 and 2004 show that desert is
color of the ground objects. To ensure high classification the main cover type across all subdivided sections in the study area.
accuracy, contingency matrixes were used to evaluate the Vegetation occurs mainly along the river in clump formation. In the
signature, and only signatures with contingency matrix value upper section, euphrate poplar forest is the major vegetation type
greater than 90% were used for classification. All images were close to the river, and the Tamarix bush occurs in zones further
classified using the maximum likelihood and supervised away, with the halophytic meadow distributed in random pattern.
classification method (Tobler et al., 2003). Accuracy assessment Halophytic meadow predominates along the old Tarim River in the
of classification was made with the 2005 field survey data on middle section; whilst in the lower section, Tamarix bush
41 geo-referenced vegetation plots and information for 40 predominates along the river, and vegetation is almost not
points from the Google Earth based on 2004 imagery of 1.0 m detectable in regions beyond 2 km of the river way. Water body
or 0.61 m resolution. The field survey plots were within 1 km2 appears only on the 2004 imagery in the river channel. Saline alkali
of the river channel along nine transects perpendicular to the soils (SAS) are mostly observable in the upper and lower sections.
river. More detailed information on those field plots is given in
Chen et al. (2006). 3.2. Land cover type
Using software FRAGSTATS 3.3 (McGarigal et al., 2002, http://www.
umass.edu/landeco/research/fragstats/fragstats.html), we calculated Different land cover types showed different patterns of change
landscape indices (Wu, 2000) for all zones in 1986, 1999, and 2004. in area between the two study periods and among the subdivided
The indices were calculated from thematic categorical map, and sections and zones (Table 1 and Fig. 1). Total vegetation area, which
included class area (CA), patch density (PD), shape index (mean value of is defined as the area with apparent presence of vegetation from
shape index of patches was used in this study) and contagion index the TM imagery analysis, decreased from 1986 to 1999 by 30%
(CONTAG). (Table 1), with the change occurring predominantly in the zone 0,
Author's personal copy

W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313 309

Table 1 3.3. NDVI


Total area (104 ha) of different land cover types for the study area in the lower
reaches of Tarim River in Xinjiang Autonomous Region, West China.
Mean NDVI varied markedly both spatially and temporally
1986 1999 Rate of change 2004 Rate of change from across the study area (Fig. 3). Spatially, it was highest in the upper
from 1986 to 1999 1999 to 2004
section and lowest in the lower section. Temporally, changes in
Desert 36.27 38.01 4.8% 34.87 8.3% mean NDVI varied with subdivision of sections and zones.
Saline alkali soil 2.52 3.14 24.6% 3.05 3.2%
In the upper section, the mean NDVI declined by 23% to nearly
Vegetation 7.86 5.49 30% 8.10 48%
Water body 0 0 – 0.62 – 50% from 1986 to 1999, and increased by 34% to as much as 300%
from 1999 to 2004. The middle section displayed similar patterns of
temporal change in the mean NDVI for all but zones E2 and E3, with
an average of decline by 25% from 1986 to 1999 and an average
E2, and E3 of the upper section, and the zone 0 of the lower section
increase by about 30%. In the lower section, the apparent decline of
(Fig. 2). From 1999 to 2004, the total vegetation area increased by
mean NDVI from 1986 to 1999 and the subsequent recovery from
48% (Table 1), with the most noticeable changes occurring in zone
1999 to 2004 only occurred along the river channel and in zones
0 for the upper section, zones 0 and W3 for the middle section, and
immediately adjacent to the river channel (Fig. 3).
zone 0 for the lower section (Fig. 2).
The desert area increased by 4.8% from 1986 to 1999 at an average
rate of 1338 ha per annum, and decreased by 8.3% from 1999 to 2004 3.4. Landscape pattern
at an average rate of 6280 ha per annum (Table 1). Increases of the
desert area from 1986 to 1999 occurred most noticeably in zones 0, 3.4.1. Patch density
E1, and E2 in the upper section, and zone 0 in the lower section; Patch density varied both spatially and temporally within the
whilst decreases of the desert area from 1999 to 2004 occurred study area, but displayed different patterns of spatial variations for
predominantly in zone 0 across the three longitudinally subdivided the three different vegetation types (Fig. 4). For all the three
sections, i.e. in the areas immediately along the river area (Fig. 1). vegetation types, the patch density generally increased from 1986
The SAS area increased by 108% from 1986 to 1999, and to 1999 and decreased from 1999 to 2004 with very few exceptions.
decreased by 14.6% from 1999 to 2004 in the upper section; in the Spatially, the euphrate poplar forest displayed an apparent
middle section, it increased by 131% from 1986 to 1999 and by tendency of decreasing patch density with distance from the river
23.3% from 1999 to 2004; whilst in the lower section, it decreased channel in both the upper and the lower sections; whilst a similar
by 25.6% from 1986 to 1999 and by 20.4% from 1999 to 2004. pattern of spatial change was observed for the Tamarix bush only in
Changes in the SAS area from 1999 to 2004 were most apparent in the upper section. The halophytic meadow was not observable in
zone 0 across all three sections, and in zone W3 of the upper and the lower section of the study area.
middle sections immediately adjacent to the old Tarim River
(Fig. 1). Overall, there was an increase in the SAS area by nearly 25% 3.4.2. Shape index
from 1986 to 1999 and then a decrease by about 3% from 1999 to Shape index was less variable spatially compared with other
2004 for the entire study area (Table 1). measures of landscape features. However, the temporal change of
Spatially, the vegetation area decreased and the desert area the shape index varied with vegetation types (Fig. 5). For euphrate
increased with distance from the river channel in the upper section poplar forest, the shape index decreased from 1986 to 1999 in all
of the study area from 1986 to 2004, while such patterns were zones with apparent exception of zone W1 in the upper section,
lacking for both the middle and the lower sections. and mostly increased from 1999 to 2004. Temporal changes in the

Upper section Middle section Lower section


100
1986
% of vegetation

80
1999
60 2004
40

20

100

80
% of desert

60

40

20
% of salt alakali soil

100

20

0
W3 W2 W1 0 E1 E2 E3 W3 W2 W1 0 E1 E2 W2 W1 0 E1 E2

Fig. 1. Percentage of vegetation, desert and saline alkali soil (SAS) over total area in individual zones in the upper, middle and lower sections of the lower reaches of Tarim River in
Xinjiang Autonomous Region, West China.
Author's personal copy

310 W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313

Upper section Middle section Lower section

Vegetation area change


6000
from 1986 to 1999
4000 from 1999 to 2004

2000

-2000

-4000
W3 W2 W1 0 E1 E2 E3 W3 W2 W1 0 E1 E2 W2 W1 0 E1 E2

Fig. 2. Area change of the vegetation type in the upper, middle, and lower sections of the lower reaches of Tarim River in Xinjiang Autonomous Region, West China, between 1986
and 1999 and between 1999 and 2004.

shape index for Tamarix bush were variable across sections and Previous field investigations showed that improved vegetation
zones, with the general pattern showing a decrease from 1986 to following the emergency water diversion was closely related to the
1999 and an increase from 1999 to 2004. The shape index for rise in groundwater table in our study area (Chen et al., 2004b,c,
halophytic meadow increased from 1986 to 1999 and decreased 2006). Precipitation in the area ranged form 3.4 to 31.2 mm per
from 1999 to 2004 in the upper section with exception of zones E2 annum during the period of water diversion, far from being
and E3; whereas there was no apparent trend of temporal change in adequate for sustaining plant growth. Elmore et al. (2006) found
the middle section. that the alkali meadow in California, USA, with average

3.4.3. Contagion index


The contagion index varied both spatially and temporally but
.16
without apparent patterns of temporal change among the three
1986 upper section
longitudinally subdivided sections along the Qiwenkuor River .14
1999
(Fig. 6). In the upper section, there was a tendency of gradually
.12 2004
increasing values of the contagion index with distance from the
river channel; in the middle section, the contagion index was .10
lowest in the zone 0 and increased abruptly in other zones without
transition; whilst there was no clear spatial pattern for the value of .08
contagion index in the lower section. The temporal change in the
.06
contagion index differed between two sides of the Qiwenkuor River
in the upper and middle sections: in zones on the east side of the .04
Qiwenkuor River channel, the contagion index decreased from
.02
1986 to 1999 and to 2004 in the upper section with exception of
zone W3 and in the middle section with exception of zone W1. The 0.00
value of the contagion index increased from 1986 to 1999 and .16
decreased from 1999 to 2004 in zone E1 of all sections and in zone Middle section
.14
E2 of the upper section, and increased from 1986 to 1999 and to
2004 in zone E3 of the upper section and zone E2 of the middle and .12
Mean NDVI

lower sections. .10

.08

.06
4. Discussion
.04
In arid and semiarid regions, vegetation and landscape patterns
.02
can be highly susceptible to both anthropogenic and environmental
perturbations. Changes in vegetation and landscape patterns can 0.00
affect plant growth and ecosystem structures and function (Chen .16
Lower section
et al., 2004b; Naumburg et al., 2005; Zhou et al., 2006, 2007; .14
Zhuang and Chen, 2006). One of the environmental perturbations,
the groundwater fluctuation, has proven a key factor in influencing .12
vegetation and landscape patterns in arid and semiarid regions .10
(Muñoz-Reinoso, 2001; Naumburg et al., 2005; Elmore et al., 2006).
In the lower reaches of the Tarim River in West China, contrasting .08
changes in groundwater level were shown between the periods .06
1986–1999 and 1999–2004 as shown in field survey (Chen et al.,
2004c). In the present study, we found that overall both vegetation .04
cover and NDVI decreased for the period 1986–1999 and increased .02
for the period 1999–2004, while the desert area displayed an
inverse pattern of change. Landscape patterns also displayed con- 0.00
W3 W2 W1 0 E1 E2 E3
trasting trends of change between the two periods, indicating the
apparent impact of groundwater fluctuation to vegetation and Fig. 3. Mean NDVI in the upper, middle, and lower sections of the lower reaches of
landscape patterns in this region. Tarim River in Xinjiang Autonomous Region, West China, for 1986, 1999, and 2004.
Author's personal copy

W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313 311

Upper section Middle section Lower section

1986

Euphrate poplar
30
1999

forest
20 2004

10
Patch density (n/100ha)

0
30
Tamarix bush

20

10

0
Halophytic meadow

30
20
10

0
W3 W2 W1 0 E1 E2 E3 W3 W2 W1 0 E1 E2 W2 W1 0 E1 E2

Fig. 4. Patch density of euphrate poplar forest, Tamarix bush and halophytic meadow in the upper, middle and lower sections of the lower reaches of Tarim River in Xinjiang
Autonomous Region, West China for 1986, 1999 and 2004.

precipitation of 130 mm per annum and snowmelt replenishment In this study, increases in vegetation cover during the period of
in spring and summer, was weakly related to precipitation only water diversion appeared to be attributable mainly to the euphrate
when groundwater was lower than 2.5 m below the surface. In this poplar forests and Tamarix bushes, and changes in the halophytic
study, the most notable change of vegetation was identified to have meadow were only observable in only a few zones. This is consis-
occurred in areas near the river channel. Percentage of vegetation tent with results from the field studies (Chen et al., 2004c).
cover decreased with distance from the Qiwenkuor River in the A previous study by Chen et al. (2004a) has shown that the func-
upper and lower sections. It is clear from results of the present tional groundwater depth is w5 m for Tamarix spp., w4.5 m for
study and those from previous field investigations (e.g. Chen et al., P. euphratica (4.5 m), and w3.5 m for P. communis. Therefore,
2004b,c, 2006) that changes in vegetation cover were closely a reversion in groundwater table by the emergency water diversion
associated with changes in groundwater level. Based on previous project would be expected to benefit Tamarix spp. first then fol-
field surveys and results in this study, two conclusions may be lowed by P. euphratica in the sequence of vegetation recovery. In
drawn: (1) that the changes in water-flow to the lower reaches of our study area, the euphrate poplar forests mainly exist as mature
the Tarim River affected the groundwater mostly in areas adjacent trees and regenerating stands are rare (Chen et al., 2005). The
to the river channel; and (2) that vegetation near the river channel observed increases in vegetation cover are thus likely to result from
are more groundwater dependent than those farther away, hence a ‘‘greening-up’’ of the recovering trees rather than an increase in
are more susceptible to groundwater fluctuations. regeneration.

Upper section Middle section Lower section


Euphrate poplar

1.6
forest

1.4
1.2
1.0
0.2

0.0
Shape index
Tamarix bush

1.6
1.4
1.2
0.2

0.0
Halophytic meadow

1.6 1986
1999
1.4 2004
1.2
0.2

0.0
W3 W2 W1 0 E1 E2 E3 W3 W2 W1 0 E1 E2 W2 W1 0 E1 E2

Fig. 5. Shape indices of euphrate poplar forest, Tamarix bush and halophytic meadow in the upper, middle and lower sections of the lower reaches of Tarim River in Xinjiang
Autonomous Region, West China, for 1986, 1999 and 2004.
Author's personal copy

312 W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313

Upper section Middle section Lower section


100
1986

Contagion index
80 1999
60 2004
40
20
0
W3 W2 W1 0 E1 E2 E3 W3 W2 W1 0 E1 E2 W2 W1 0 E1 E2

Fig. 6. Contagion index for the upper, middle and lower sections of the lower reaches of Tarim River in Xinjiang Autonomous Region, West China, in 1986, 1999 and 2004.

Temporally, mean NDVI decreased in all zones from 1986 to limited by resources and this made it easily influenced by envi-
1999, indicating the dependency of vegetation on groundwater. The ronmental perturbation. Changes in landscape shape meant that
change was different among the three sections, possibly related to the patch shape was influenced more by environment than their
the difference of cover type and lack of effective groundwater individual characteristics. Turner (1989) noted that the shape of
recovery downstream. Spatially, mean NDVI decreased from the landscape influenced by natural rather than anthropogenic
upper section downward to the lower section, displaying apparent disturbances may respond differently, with natural disturbances
trend with distance from the river channel in each of the sections increasing landscape complexity. Our study showed, however, that
except only a few zones (Fig. 3). Zones with higher NDVI away from the same factors that operate in different directions could lead to
the river channel generally had meadow components as shown on either increased or decreased landscape complexity.
the vegetation classification map. The values of NDVI reflect the Management strategies for restoring groundwater level are
extent of coverage for ground vegetation as well as the vigor of necessary for conserving plant resources in arid regions. Results
vegetation growth. Meadows in this area are usually clumped and from this study indicate strongly the hydrological control of spatio-
therefore are shown with higher coverage than forests and shrubs, temporal variations of vegetation and landscape pattern in arid
which could explain the irregular distribution of the mean NDVI regions. Our overall findings suggest that water diversion to
farther away from the river channel. Moreover, those zones are also previously drained water channel is effective for raising the local
special zones in the water division. Zones E2 and E3 in the upper groundwater level and improving plant growth, but its effect is
section are irrigated as part of water diversion plan to restore largely restricted to areas adjacent to the water pathway. For long-
vegetation. Zone W3 in the middle section overlaps with the old term conservation of plant resources in arid regions, conserving
Tarim River, which was also the water delivery pathway in the fifth water usage and keeping the source of groundwater supply intact
diversion. should be taken as preferred management options. Temporary
Compared with the conventional ground-based surveys, remote measures such as the emergency water diversion project in the
sensing imagery analysis offers an effective tool for assessing lower reaches of the Tarim River should only be used where long-
vegetation at a much broader spatial scale. Our results showed that term conservation of plant resources is socially justifiable, and
the emergency water diversion from 2000 to 2004 has a much economically and environmentally sustainable.
wider effect spatially than the ground-based investigations (Chen
et al., 2004c) and the effects differed with sections and zones. Acknowledgement
Vegetation in zones outside of the groundwater-affected area
decreased from 1986 to 1999 and to 2004. A general trend of This study was supported by the Knowledge Innovation Project
decreasing precipitation may have caused the declining vegetation of the Chinese Academy of Sciences (90502004) and the National
far from the river system. Natural Science Foundation of China (30500081). We thank Peter
Along with changes in vegetation, landscape features during the Barry, T. Wang, and Q. Li for technical help with using the software
two study periods also changed. From 1986 to 1999, vegetation FRAGSTATS, and Professor J. Ni for supplying meteorological data.
tended to be fragmented as indicated by results of patch density;
shape of vegetation patches tended to be regular as shown by shape
index; and landscape patches distributed more unevenly, as shown References
by contagion index. In contrast, the changes from 1999 to 2004
Anyamba, A., Tucker, C.J., 2005. Analysis of Sahelian vegetation dynamics using
occurred in the opposite directions: vegetation tended to be inte- NOAA-AVHRR NDVI data from 1981–2003. Journal of Arid Environments 63,
grative; shape of vegetation patches tended to be irregular; and 596–614.
landscape patches distributed more evenly. Chen, Y.J., Chen, Y.N., Li, W.H., 2005. Change of groundwater quality and restoration
of the forests of Populus euphratica under transfusing stream water to the lower
Results on the shape change of patches revealed a very reaches of the Tarim River. Arid Zone Research 22, 101–105.
complicated picture. The overall trend of change suggested that the Chen, Y.N., Cui, W.C., Li, W.H., Zhang, Y.M., 2003a. Utilization of water resources and
shape of patches became regular from 1986 to 1999 and irregular ecological protection in the Tarim River. Acta Geography Sinica 58, 215–222.
Chen, Y.N., Li, W.H., Xu, H.L., Liu, J.Z., Zhang, H.F., Chen, Y.P., 2003b. The influence of
from 1999 to 2004. Among the three vegetation types halophytic groundwater on vegetation in the lower reaches of Tarim River, China. Acta
meadow changed differently from the euphrate poplar forests and Geogrophica Sinica 58, 542–549.
the Tamarix bush in the upper section. The shape of vegetation Chen, Y.N., Li, W.H., Chen, Y.P., Zhang, H.F., Zhuang, L., 2004a. Physiological response
of natural plants to the change of groundwater level in the lower reaches of
patches changed heterogeneously and had no apparent spatial Tarim River, Xinjiang. Progress in Natural Science 14, 975–983.
change tendency. Previous studies have generally focused on the Chen, Y.N., Wang, Q., Ruan, X., Li, W.H., Chen, Y.P., 2004b. Physiological response of
impact of anthropogenic activity on landscape shape and the Populus euphratica to artificial water-recharge of the lower reaches of Tarim
River. Journal of Integrative Plant Biology 46, 1393–1401.
available results have shown that the shape of human-disturbed
Chen, Y.N., Zhang, X.L., Zhu, X.M., Li, W.H., Zhang, Y.M., Xu, H.L., Zhang, H.F.,
landscape is more regular than natural landscape, but differed Chen, Y.P., 2004c. Analysis on the ecological benefits of the stream water
among disturbance activity (plantation or farming, etc.) (Saura and conveyance to the dried-up river of the lower reaches of Tarim River, China.
Carballel, 2004). Our study showed that under the declining Science in China Series D – Earth Sciences 47, 1053–1064.
Chen, Y.N., Zilliacus, H., Li, W.H., Zhang, H.F., Chen, Y.P., 2006. Groundwater level
groundwater availability, the shape of vegetation patches tended to affects plants species diversity along the lower reaches of the Tarim River,
be regular. This may be because plants on the patch margin are western China. Journal of Arid Environments 66, 231–246.
Author's personal copy

W. Kong et al. / Journal of Arid Environments 73 (2009) 306–313 313

Elliott, L.J., Mason, D.C., Wilkinson, M.J., Allainguillaume, J., Norris, C., Alexander, M., District, Kenya. ISPRS Journal for Photogrammetry and Remote Sensing 59,
Welters, R., 2004. The role of satellite image-processing for national-scale 103–114.
estimates of gene flow from genetically modified crops: rapeseed in the UK as Tobler, M.W., Cochard, R., Edwards, P.J., 2003. The impact of cattle ranching on
a model. Journal of Applied Ecology 41, 1174–1184. large-scale vegetation patterns in a coastal savanna in Tanzania. Journal of
Elmore, A.J., Manning, S.J., Mustard, J.F., Craine, J.M., 2006. Decline in alkali meadow Applied Ecology 40, 430–444.
vegetation cover in California: the effects of groundwater extraction and Turner, M.G., 1989. Landscape ecology: the effect of pattern on process. Annual
drought. Journal of Applied Ecology 43, 770–779. Revolution Ecological System 20, 171–197.
Gilabert, M.A., González-Piaueras, P.J., Garcı́a-Haro, F.H., Meliá, J., 2002. A general- Weiss, J.L., Gutzler, D.S., Allred-Coonrod, J.E., Dahm, C.N., 2004. Long-term vegeta-
ized soil-adjusted vegetation index. Remote Sensing of Environment 82, tion monitoring with NDVI in a diverse semi-arid setting, central New Mexico,
303–310. USA. Journal of Arid Environments 58, 249–272.
Li, X., Lu, L., Cheng, G.D., Xiao, H.L., 2001. Quantifying landscape structure of the Wolter, P.T., White, M.A., 2002. Recent forest cover type transitions and
Heihe River Basin, north-west china using FRAGSTATS. Journal of Arid landscape structural changes in northeast Minnesota, USA. Landscape Ecology 17,
Environments 48, 521–535. 133–155.
Lillesand, T.M., Kiefer, R.W., 2000. Remote Sensing and Image Interpretation, fourth Wu, J.G., 2000. Landscape Ecology: Pattern, Process, Scale and Hierarchy. Higher
ed. John Wiley & Sons, New York. Education Press, Beijing, pp. 100–109.
McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: Spatial Pattern Wu, J.G., Hobbs, R., 2002. Key issues and research priorities in landscape ecology: an
Analysis Program for Categorical Maps. Computer software program. The idiosyncratic synthesis. Landscape Ecology 17, 355–365.
University of Massachusetts, Amherst. Available from:. www.umass.edu/ Wu, J.G., Shen, W.J., Sun, W.Z., Tueller, P.T., 2002. Empirical patterns of the effects of
landeco/research/fragstats/fragstats.html. changing scale on landscape metrics. Landscape Ecology 17, 761–782.
Muñoz-Reinoso, J.C., 2001. Vegetation changes and groundwater abstraction in SW Xu, H.L., Mao, Y., Li, J.M., 2007. Changes in groundwater levels and the response of
Doñana, Spain. Journal of Hydrology 242, 197–209. natural vegetation to transfer of water to the lower reaches of the Tarim River.
Naumburg, E., Mata-Gonzalez, R., Hunter, R.G., Mclendon, T., Martin, D.W., 2005. Journal of Environmental Sciences 19, 1199–1207.
Phreatophytic vegetation and groundwater fluctuations: a review of current Zha, Y., Gao, J., Ni, S.X., Liu, Y.S., Jiang, J.J., Wei, Y.C., 2003. A spectral reflectance-
research and application of ecosystem response modeling with an emphasis on based approach to quantification of grassland cover from Landsat TM imagery.
Great Basin vegetation. Environmental Management 35, 726–740. Remote Sensing of Environment 87, 371–375.
Palmer, A.R., Van Rooyen, A.F., 1998. Detecting vegetation change in the southern Zhang, Y.M., Chen, Y., 2004. Plant communities and its interrelation with environ-
Kalahari using Landsat TM data. Journal of Arid Environments 39, 143–153. mental factors in the lower reaches of Tarim River valley. Acta Geographica
Saura, S.S., Carballel, P., 2004. Discrimination of native and exotic forest patterns Sinica 59, 903–910.
through forest irregularity indices: an analysis in the landscape of Galicia, Zhou, Z.Y., Sun, O.J., Huang, J.H., Gao, Y.Z., Han, X.G., 2006. Land use affects the
Spain. Landscape Ecology 19, 647–662. relationship between species diversity and productivity at the local scale in
Song, Y.D., 1999. Research on Water Resources and Ecology of Tarim River, China. a semi-arid steppe ecosystem. Functional Ecology 20, 753–762.
Xinjiang People’s Press, Urumqi, pp. 216–244. Zhou, Z.Y., Sun, O.J., Huang, J.H., Li, L.H., Liu, P., Han, X.G., 2007. Soil carbon and
Stromberg, J., Tiller, R., Richter, B., 1996. Effects of groundwater decline on riparian nitrogen stores and storage potential as affected by land-use in an agro-pastoral
vegetation of semiarid regions: the San Pedro, Arizona. Ecological Applications ecotone of northern China. Biogeochemistry 82, 127–138.
6, 113–131. Zhuang, L., Chen, Y.N., 2006. Physiological responses of three contrasting plant
Tachiiri, K., 2005. Calculating NDVI for NOAA/AVHRR data after atmospheric species to groundwater level changes in an arid environment. Journal of
correction for extensive images using 6S code: a case study in the Marsabit Integrative Plant Biology 48, 520–526.

Оценить