Botanical Studies (2008) 49: 57-66.

*

Corresponding author: E-mail: gejp@bnu.edu.cn;

Fax: +086-010-58808999.

INTRODUCTION

As an important component of the terrestrial ecosystem,

vegetation has long been a research focus for ecologists

and botanists. The role of vegetation in soil and water

conservation on the Loess Plateau has been well docu-

mented (Chen and Wan, 2002). However, how seriously

the current vegetation has been degraded and how far it

is from its original state are hotly debated. According to

geobotany and historic geography, human activities are

responsible for the degradation of the original vegetation

in the Loess Plateau and have led to a big gap between

the current vegetation and the original vegetation (Wu,

1980; Zhu, 1983; Shi, 1991). Physical geographists and

quaternary environmentalists have argued that the gap

between the current vegetation and the original vegetation

is not as big as some have asserted, and that the Loess

Plateau was not within the typical forest distribution

region from the perspective of climate and geological

history (Zhang, 1992; Liu et al., 1996; 2003). The origi-

nal vegetation is the undisputed key to this controversial

question, but this no longer exists in most of the Loess

Plateau.

A number of small-scale local studies of the vegetation

degradation on the Loess Plateau have been carried

out (Chen and Wan, 2002), but landscape and regional

scale investigations with adequate ecological details are

still scarce. Such large scale studies have now become

feasible largely because of the development of remote

sensing and spatial information technologies (Wu and

Hobbs, 2002). Remote sensing has been widely used to

monitor vegetation dynamics (Bastin et al., 1995; Tong

et al., 1996; Boyle et al., 1997; Tanser and Palmer, 1999;

Schmidt and Karnieli, 2000). Many, if not most, of these

studies, however, evaluated the conditions of vegetation

degradation based only on an above-ground biomass

estimated using the NDVI. Potential vegetation is defined

as the maximum biomass which would be established in

a given area assuming no human influence (Rey, 1999).

The term potential evapotranspiration (PET) is needed

to determine the hydrological balance, which in turn

dictates the potential vegetation that a particular place

may harbor (Woodward, 1987; Running and Coughlan,

1988; Woodward and Mckee, 1991; Prentice et al., 1992;

Neilson, 1993; Prentice et al., 1993; Neilson and Marks,

1994; Neilson, 1995). Most studies to date have only

focused on PET modeling and assumed that large-scale

vegetation units can be characterized by the predominance

of one or more PETs (Rey, 1999; Zhao, 2004; Kimura

et al., 2005). None has yet taken quantitative transient

responses of potent vegetation to continuing changes of

PET into account.

The major objective of this study was to accomplish a

more detailed regional assessment of vegetation deficiency

in Jinghe River Basin¡Xone of the most representative

Vegetation deficiency in a typical region of the Loess

Plateau in China

Anning SUO, Yong LIN, Jianping GE*, and Xiaojun KOU

College of Life Science, Beijing Normal University, Beijing 100875, P.R. China

(Received February 21, 2006; Accepted August 6, 2007)

ABSTRACT.

Located in the central Chinese Loess Plateau, the Jinghe River is one of the major tributaries

of the Yellow River, and the river basin it inhabits is characterized by a continental climate. Considering that

land degradation has been the main ecological problem in the Jinghe River Basin, there is an urgent need to

scientifically explore the land degradation mechanism in order to restore natural ecological environments,

prevent further irreversible degradation, and retain value of these productive lands. To that end, a potential

vegetation prediction and current vegetation degradation evaluation of the study area are the first and foremost

jobs worth doing. In this paper, the Holdrige method is used to calculate the annual Aridity Index (AI), and

a climate-vegetation based AI-NDVI model was proposed to predict the potential vegetation index, on which

basis the potential vegetation index and potential vegetation distribution were both obtained. With the potential

vegetation index and current vegetation index, the gap between the current and the potential vegetation was

easy to calculate. This gap exhibits vegetation degradation objectively in the Jinghe River Basin. The AI-

NDVI approach is expected to prove effective for ecological planning and management on a landscape and

regional scale and to fuel integration between hydrology and ecology during ecological restoration research.

Keywords: Ecological restoration; Potential vegetation; The Loess Plateau; Vegetation deficient index.

eCOLOgy

58

Botanical Studies, Vol. 49, 2008

geographic areas on the Loess Plateau. Our approach was

to combine remote satellite data (MODIS), climatic data,

DEM data, and additional field surveys. We were also

interested in developing a synoptic index of vegetation

deficiency that can be used to quantify vegetation defi-

ciency in terms of both its spatial distribution and temporal

dynamics. By assessing the spatial pattern and temporal

dynamics of the vegetation deficiency in the Jinghe

River Basin, we intended the project to provide a useful

approach for large-scale assessment and management of

the vegetation resources on the Loess Plateau.

STUDy SITe AND DATA SOURCe

Study site

Located in the central part of the Chinese Loess

Plateau, the Jinghe River Basin (with a total land area of

45,420 km

2

) drains into the middle reach of the Yellow

River (Figure 1). The terrain is undulating with elevations

ranging from 800 to 2,600 m. The climate is characterized

by cold, dry winters and warm, humid summers. The

climatic conditions in the study area show obvious spatial

variation. The mean annual air temperature is 8.8¢XC at

the northern end of the basin and increases to 10¢XC at the

southern end of the basin. The mean annual precipitation

is more than 600 mm at the western and eastern mountain

side, gradually decreasing northward to only 300 mm in

the northern end. There are three special characteristics of

the precipitation. First, the its inter-annual variability is

as high as 35%. Second, over 60% of it falls between July

and September. Third, over 40% of it falls during storms.

The hilly topography and intensive precipitation allows

only a small portion of the precipitation to recharge the

soil water supporting plants. The major portion runs off

into rivers during storms. In addition, extensive human

removal of the natural vegetation and farming-induced

Figure 1. The location of Jinghe River Basin. Note: 1, Huajia; 2, Dongchuan; 3, Puhe; 4, Ruhe; 5, Heshui; 6, Dongzhi; 7, Honghe; 8,

Upperinghe; 9, Ruihe; 10, Chengguhe; 11, Heihe; 12, Daxihe; 13, Sanshuihe; 14, Lowjinghe.

SUO et al. ¡X Vegetation deficiency in a typical region of the Loess Plateau in China

59

hydrological deterioration of the surface soil further

exacerbates the hydrological condition of Jinghe River

Basin. Thus, that the area studied contributes a solid load

to the Yellow River ten times above expectations should

be no surprise.

The basin has a long history of agricultural

development. Most river valleys and loess gullies in

the basin have been extensively cultivated, and only a

small amount of secondary vegetation remains at the

gully bottoms and on the ridges. More than 1320 plant

species belonging to 597 genera and 128 families inhabit

the basin, and these include Bryophyta, Pteridophyta,

Gymnospermae and Angiosperm. Herbaceous plants

and trees account for more than 900 and 400 of these

species, respectively. The dominant families are Fagaceae,

Betulaceae, Salicaceae, Rosaceae, Pinaceae, Poaceae,

Asteraceae. On the horizontal zone from the south basin to

the north basin, sparse forests mixed with arid shrublands

(composed of Ulmus pumila,

Sophora japonica, Malus

baccata),

arid shrublands (dominated by Ziziphus jujuba

var. spinosa,

Vitex negundovar. heterophylla,

Sophora

viciifolia and Bothriochloa ischaemum), grasslands

(with Leymus paboanus, Poa annua and Roegneria

nutans as constructing species) and desert grasslands

(composed mainly of Stipa bungeana and Stipa breviflora )

are distributed in order. In the mountainous parts of

Liupanshan and Zhiwuling, arid shrublands (mainly

composed of Ostryopsis davidiana, Spiraea pubescens,

Spiraea trilobata and Hippophae rhamnoides) and broad-

leaf forests of Populus davidiana, Betula platyphylla, and

Quercus liaotungensis

are distributed in vertical order

from bottom to top (Chen and Wan, 2002).

Data sources

Remote sensed data (spanning from 2001 January to

2004 December) used in this study are a MODIS 32-day

product at 500 m resolution provided by the Global Land

Cover Facility (GLCF). For the convenience of analysis,

the date has changed from the Julian day to a calendar day

in advance.

Monthly average temperatures from 30 stations and

precipitation from 95 rain-gauge stations in the Jinghe

River Basin and its adjacent regions were provided by the

China Lanzhou Arid Agriculture Climate Bureau, China¡¦s

climatic bureau.

The Jinghe River topographic Map at scale of 1:250,000

is from the Institute of Geography and Resources, Chinese

Academy of Sciences.

MATeRIALS AND MeTHODS

Arid index calculation

Many studies have proved that vegetation distribution

is closely related to climate (Fang and Yoda, 1990;

Zeng et al., 1996; Richard and Poccard, 1998; Fu and

Wen, 1999). In the Loess Plateau water availability is

a decidedly limiting factor affecting plant distribution

and is usually expressed by the aridity index (AI) to

indicate climatic aridity. AI can be calculated by the

ratio of water balance to heat balance (Wang and Xiao,

1993). Many aridity indexes have been put forth since

1990, such as the ratio of temperature to rainfall and the

ratio of potential evapotranspiration (PET) to rainfall.

Zhang (1993) employed the Holdrige method to plot a

China PET map on the basis of 700 climatic stations with

a more satisfactory vegetation-climate classification,

which showed that the Holdrige method is a convenient

quantitative analysis method. Other studies also found

that the Holdrige method is relatively precise and agrees

well with vegetation distribution (Zhou and Zhou, 1996).

Therefore, the Holdrige method is used to calculate AI in

this paper.

In the Holdrige method, potential evapotranspiration

(PET) is calculated by an empirical formula with

temperature as an input variable (Holdrige, 1947, 1976).

The formula is as follows:

PET=58.93¡ÑABT

(1)

ABT=£Ut/365

(2)

ABT=£UT/12

(3)

where PET is potential evapotranspiration (mm), ABT

is annual bio-temperature ranging from 0 to 30¢XC. Daily

temperatures above 30¢XC are set to 30¢XC, and those below

1¢XC are set to 0¢XC.

The ratio of PET to rainfall (P) is used to express the

aridity index, and the formula is as follows:

K = PET/P

(4)

where K is the aridity index, and P is average annual

rainfall.

Monthly climatic data from 1970 to 2003 was

first processed to get the average annual rainfall for

each climatic station. Then we used the DEM derived

from the 1:250,000 topographic map to obtain the

temperature of each station at sea level by means of a

lapse rate 0.65¢XC/100 meter and 100¡Ñ100 m grid maps

of monthly temperature by way of spatial interpolation.

The temperature of each cell is revised to get an actual

temperature based on its corresponding elevation and the

temperature lapse. The spatial distribution of potential

evapotranspiration was then derived based on the

temperature map.

Research showed that precipitation is closely related

to longitude, latitude, and elevation (Zhao, 1988). In

this paper, annual precipitation is estimated by latitude,

longitude, and elevation on the basis of 1970~2003

precipitation data. The regression function is as follows:

P =

1297.634 - 165.413LAT + 46.822LONG +

0.0815ALT

(5)

R

2

=0.769, a = 0.01, n =32

with the support of GIS, a grid map of precipitation

is derived based on geographic position and the

corresponding elevation of the rain gauge station and

60

Botanical Studies, Vol. 49, 2008

the regression function (5). Finally, we got the Holdrige

aridity index distribution map of the Jinghe River Basin

with the formula (4) (Figure 2).

Vegetation index calculation

The Normalized Difference Vegetation Index (NDVI)

is an important monitoring index of surface vegetation

features. NDVI is related to biomass, leaf area index,

photosynthesis capacity, total dry mass accumulation,

and net primitive yield (Birky, 2001). Therefore, NDVI

is commonly used to assess vegetation cover and its

dynamics monitoring on a large scale and simulate bio-

physical parameters for various vegetation, land cover,

and climatic change research (Fuller, 1998). NDVI can be

calculated by the following formula:

NDVI = (£l

ch2

¡V£l

ch1

)/(£l

ch2

+£l

ch1

) NDVI = (-1, 1) (6)

where £l

ch1

and £l

ch2

are the reflection of band 1(ch1) and

band 2 (ch2). Ch1 (0.58~0.68 £gm) belongs to the visible

band and is within the absorption band of green color, and

ch2 (0.75~1.10 £gm) is the infra-red band and within green

plant¡¦s reflection band.

MODIS image preprocessing, NDVI computing, and

false color composition were carried out with ERDAS

8.7 software to get average monthly NDVI from January

to December and average summer NDVI (June, July, and

August) for three years.

In this study, Jinghe River Basin on the Loess

Plateau was chosen as a study region, and GIS-assisted

methods were employed to estimate the spatial-temporal

distribution of vegetation degradation by the following

procedures. First, the Holdrige¡¦s method was used to

estimate PET and calculate aridity index (AI). Second, a

regression model of an aridity index (AI) and a normalized

difference vegetation index (NDVI) was employed to

model a potential vegetation index. Third, the vegetation

deficient index expressed by the gap between potential

vegetation and current vegetation was modeled spatially to

show the vegetation deficient condition.

ReSULTS

Relationship between aridity index and

vegetation index

Spatial distribution of climatic aridity index reflects

the pattern of water-heat balance (Kimura et al., 2005). In

the Jinghe River Basin (Figure 2), the aridity index was

the lowest (AI<0.80) on Ziwuling Mountain in the eastern

basin and Liupanshan Mountain in the western basin. It

ranged from 0.80 to 0.90 in southern mountain region.

Because rainfall increases and temperature decreases

with elevation, the water-heat balance was good in the

mountain region. The aridity index in the low plain was

much higher than in its neighboring loess gully regions,

reflecting the important influence of local topography on

regional water-heat balance. The aridity index gradually

became worse with rainfall decreasing from south to

north in the whole basin. The aridity index was 0.8 on the

southern mountains, 1.0 in the region¡¦s center, and above

1.50 north of the basin.

As we all know, the distribution of natural vegetation

is significantly related to the climatic aridity index. So 50

points of natural vegetation with relatively small human

disturbance were selected, and their corresponding AI

and NDVI values were calculated (Figure 2 and Figure

3).

With these data, a scatter plot of NDVI-LAI is shown

(Figure 4). It is clear that NDVI was negatively correlated

with AI.

Using logic and multi-level models based on

the statistical software Statistica 6.0, equations and

relationships can be developed, which contain variables

and their appropriate weights for spatially-explicit

modeling using linear regression analyses approaches. A

linear regression relationship between NDVI and AI with

the best significance was selected:

NDVI = 1.647EXP (-1.039AI)

(7)

R squared = 1-Residual SS/Corrected SS = 0.863,

a = 0.01, n = 50

where AI is the aridity index, NDVI is the normal

different vegetation index, R squared is the correlation

coefficient, residual SS is the sum of squares of residual,

and corrected SS is the sum of squares of corrected total,

a is the significant range, and n is the sample number.

This tells us that R squared equals 0.863, which is the best

performance of the model and makes the model the best

choice.

Figure 2. Aridity index spatial pattern in Jinghe River Basin.

SUO et al. ¡X Vegetation deficiency in a typical region of the Loess Plateau in China

61

Spatial pattern of potential vegetation index

The Potential Normalized Difference Vegetation Index

(PNDVI) of the Jinghe River Basin is obtained based

on the formula (7) with the support of Arc/info GIS.

PNDVI in this basin mainly ranged from 0.50 to 0.70,

which accounts for 56.15% of the basin¡¦s area, in contrast

to 31.28% (PNDVI>0.70) and 12.57% (PNDVI<0.50)

(Figure 5).

The spatial pattern of PNDVI varied from watershed

to watershed of the basin. It was highest in the southern

part of the basin and lowest in the northern part of the

basin, showing a rising trend from north to south. The

Daxihe watershed, in the south mountain region of the

basin, boasts a 53.80% land area with PNDVI>0.80 and a

96.76% land area with PNDVI>0.70. In contrast, for the

Honghe watershed, located in the basin¡¦s central part, the

land area with PNDVI>0.70 only accounts for 13.7% of

the watershed while land with a PNDVI ranging from 0.60

to 0.70 accounts for 81.56% area of the watershed. The

PNDVI is smaller in general in the Huajia watershed in the

northern basin. The PNDVI in its 52.47% land area is less

than 0.50, and that in its 95.67% land area is lower than

0.60.

The PNDVI is higher in the mountainous regions of the

western and eastern side of the basin and lower in the Yuan

region of the central basin. The PNDVI in 65.39% of the

Sanshuihe watershed in Ziwuling Mountain exceeds 0.80.

The land with a PNDVI bigger than 0.80 in the Ruihe

watershed of the west Liupanshan Mountain accounts for

42.70%. The PNDVI in the Dongzhi region on central

loess Yuan and loess gully region was lower with only

3.42% of the watershed boasting a PNDVI >0.70.

Spatial pattern of potential vegetation types

Six main natural vegetation types, including warm-

temperate deciduous broad-leaved forest, coniferous and

broad-leaved mixed sparse forest, mesophytic shrub, dry

shrub, meadow and temperate steppe, are distributed in

the Jinghe River Basin according to field investigations

and research reports (Chen and Wan, 2002). The potential

NDVI image of the basin was classified into six natural

vegetation types by a unsupervised classification approach

and referred to 50 selected points (Figure 6).

Figure 3. Vegetation index (NDVI) spatial pattern in Jinghe

River Basin.

Figure 4. Relationships between vegetation index and aridity

index.

Figure 5. Potential vegetation index spatial pattern in Jinghe

River Basin.

62

Botanical Studies, Vol. 49, 2008

As for the whole basin, the dominant vegetation types

in terms of area ratio are warm-temperate deciduous

broad-leaved forests (31.28%), coniferous and broad-

leaved mixed sparse forests (32.44%), and mesophytic

shrubs (23.71%). Warm-temperate deciduous broad-leaved

forests are mainly composed of Populus davidiana, Betula

platyphylla, Quercus liaotungensis et al., and mainly

distributed on Ziwuling mountain in the eastern basin,

Liupanshan Mountain in the western basin and southern

mountain of the basin. Coniferous and broad-leaved mixed

sparse forests with Betula platyphylla suk, Ulmus pumila,

Sophoro japonica, and Malus baccata as constructed tree

species are located in the watersheds of Honghe, Ruhe,

low reaches of the Jinghe, Ruihe, Guchenghe watersheds

and the Dongzhi region in the central basin. Ostryopsis

davidiana, Spiraea pubescens, S. trilobata and other

mesophytic shrubs are mainly distributed in loess hilly

regions of Puhe, Dongchuan, and the low reaches of

Huajia. In addition, arid mixed shrubs of Ziziphus jujuba

Mill. var. spinosa, Vitex negundo var. heterophylla, and

Sophora davidii exist in the valley and loess hilly middle

regions of Huajia. Temperate steppe of Leymus paboanus,

Poa annua, Roegneria nutans, congregated in the upper

reaches of Huajia in the northern most sections of the

basin. In all, 87.43% of the mountain and loess gully

region in middle and south of the basin with potential

vegetation was forest and shrub, and only 12.57% of the

northern basin was a steppe potential distribution region.

Spatial pattern of vegetation deficiency

We defined the Vegetation Deficient Index (VDI) as

the gap between current vegetation index (NDVI) and

potential vegetation index (PNDVI). The spatial pattern of

the VDI is displayed in Figure 7. Vegetation in the Jinghe

River Basin was seriously damaged, which is indicated

from the fact that 96.30% area of the whole basin had

vegetation of different degrees of deficiency. Vegetation,

in the rank of most deficiencies (VDI<-0.40), is noted as

very serious (-0.40 ~ -0.30) or serious (-0.30 ~ -0.20),

accounting for 2.05%, 13.97%, and 35.70% of the whole

basin area, respectively. In addition, 44.55% of the basin

(34.15% with VDI ranging from -0.20 to -0.10 and 10.40%

with VDI at range of -0.10~0) has a slightly deficient

status.

Vegetation deficiency is affected by topography,

climate, human activity, and other factors. Vegetation

of an extremely deficient status is mainly distributed

in watersheds of Ruhe and Honghe in the loess gully

region northwest of the basin, which can be seen from

the following figures: VDI<-0.30 (29.85%), VDI<-0.20

(87.64%) in the Honghe watershed, VDI < -0.30 (61.20%)

and VDI<-0.20 (91.33%) in the Ruhe watershed.

Vegetation in the Ziwuling Mountain region in the

eastern basin was slightly deficient. 29.24% of the Heshui

watershed area on Ziwuling Mountain had no vegetation

deficiency, and 26.66% was only slightly deficient.

56.40% of the Sanshuihe watershed area on Ziwuling

Mountain had a VDI of more than -0.10, and 30.11% had

Figure 6. Potential vegetation spatial pattern in Jinghe River

Basin.

Figure 7. Vegetation deficient index (VDI) spatial pattern in

Jinghe River Basin.

SUO et al. ¡X Vegetation deficiency in a typical region of the Loess Plateau in China

63

a VDI ranging between -0.20 ~ -0.10. Because human

disturbance in the Huajia watershed in the northern basin

is slight thanks to its arid climate, complex topography,

and small population, the vegetation in this watershed

is not as seriously deficient as we had thought. 59.72%

of the watershed had a VDI ranging from -0.20 to -0.10,

and 31.22% ranged from -0.30 to -0.20. Due to its flat

topography and dense population, most of the land in

Jingheganliu is cultivated for farming and other land

usage, so vegetation deficiency in this region is much more

serious than other watersheds. 7.22% of the region is with

the most serious vegetation deficieny, and 19.33% of the

area had a very serious vegetation deficiency, and 24.14%

had a serious vegetation deficiency.

Temporal pattern of vegetation degeneration

Typical points of the six main land cover types in the

basin were selected to obtain a current NDVI value and

their corresponding potential NDVI. We then picked out

potential NDVI dynamics of different land cover types

from current NDVI series with points that their average

summer NDVI value equal to PNDVI value and made

vegetation deficient dynamics in a year (Figure 8). The

NDVI of forests coincides very well with its PNDVI, and

its VDI dynamics always varied near 0 in a time series

with no vegetation deficiency. The vegetation deficiency

of shrublands was small fron January to spring, and it

increased from May, peaking (VDI=-0.20) in June, and

varied in the range of -0.20 ~ -0.15 later. Deficiency of

meadow and pasture (gaps between their current NDVI

and PNDVI) was big throughout the year, especially,

from June to September. Its VDI was -0.17 in July, -0.25

in August, and -0.28 in September. VDI of pasture was

-0.22 in June, -0.17 in July, -0.21 in August and -0.25

in September. Crops are one of the main manmade land

covers, and its vegetation deficiency showed significant

human disturbance. The VDI was near 0 from January to

spring. It decreased to -0.24 in May, 0.43 in June, 0.51

in July and 0.42 in August, with winter wheat ripening in

May and being harvested in June and the crop land resting

in July and August. The next year winter wheat is planted

and germinated in September and developed in October,

so its VDI increased to -0.39 in September and -0.08 in

October. Desert grasslands were one of the main land

cover types in the northern part of the basin. Its current

NDVI was only 0.30, and its VDI was -0.18 compared to

its PNDVI in the summer, which stayed more than -0.15 in

later months.

DISCUSSION AND CONCLUSION

Evaluating a vegetation deficient condition objectively

is difficult because no objective vegetation evaluation

criterion exists. Many researchers have suggested using

natural vegetation, original vegetation, or potential

vegetation to evaluate a vegetation degeneration situation

(Hou, 1983; Prentice et al., 1992; Yu et al., 2000; Sun and

Chen, 1991; Ren, 2004). However, natural vegetation has

only existed locally and sporadically on the Loess Plateau

for hundreds of years¡¦ cultivation in most places. The

original vegetation does not exist at all. Paleoecological

reconstruction is a main approach in original vegetation

studies. One problem is that the creation of a detailed

original vegetation distribution criterion using point

paleoecological remnants is difficult. Potential vegetation

was a credible criterion in vegetation evaluation and

ecological rebuilding. However, there is still no systematic

study on potential vegetation and no referent potential

vegetation map on the Loess Plateau (Fang, 2001; Li et al.,

1998). The gap between current vegetation and potential

vegetation is a very important research topic in ecological

planning and restoration on the Loess Plateau. It could tell

us how serious the current vegetation has been degraded.

Therefore, potential vegetation is an ideal vegetation

background in degraded vegetation evaluation.

In this paper, we first simulated the potential vegetation

distribution in the Jinghe River Basin. Then, we evaluated

spatio-temporal pattern of vegetation deficiency. The

results showed: (1) the normalized difference vegetation

index (NDVI) is closely related to the climatic aridity

index (AI), and their regression model was NDVI =

1.647EXP (-1.039AI). (2) The potential vegetation in

the mountainous region of the western basin, the eastern

basin and most of the southern basin was warm-temperate

deciduous broad-leaved forest while coniferous and broad-

leaved mixed sparse forest, and mesophytic shrub were the

potential vegetation in the loess gully region in the central

basin. Potential vegetation in the northern part of the

basin included dry shrub, meadow, and temperate steppe.

In total, 87.43% of the middle-southern basin is a forest

and shrub potential distribution region, and only 12.57%

is a grass potential region. (3) Vegetation deficiency was

high in the watersheds of Puhe, Ruhe, Honghe and lower

reaches of Jinghe, Ruihe, Heihe and Jingheganliu, but

very low on Ziwuling Mountain in the eastern basin, in the

Daxihe watershed on the southern mountain, on the upper

land of Jinghe, and on Ruihe on Liupanshan Mountain

Figure 8. Dynamic of vegetation deficient index in a year.

64

Botanical Studies, Vol. 49, 2008

in the western basin. It was moderate in the Loess hilly

region in the northern basin. (4) In the time series, forests

were not significantly deficient in all months of the one

year and desert grass lands had a smaller deficiency.

In contrast, the VDI for the winter wheat crop was the

biggest, especially from June to September when the

winter wheat was harvested in the crop land.

Experimental evidence on the potential vegetation and

on the current vegetation relative deficiency is lacking.

Such evidence would require a lengthy and costly

investigation. However, the maps shown in Figure 5 and

Figure 6 are consistent with the distribution of climate and

natural vegetation of the basin, and they constitute the best

possible validation of the proposed method (Vegetation

Atlas of China, 2001). In spite of oversimplification of

the algorithms or parameters, the method demonstrates

that vegetation deficiency is markedly consistent with that

surveyed. The present method may be considered to be

a more useful and representative vegetation assessment,

qualities which are scarce in most previous assessments

(Wu and Hobbs, 2002). The method in this paper is

proposed as an initial approximation to resolve the

deficiencies of current vegetation assessments in most

degraded vegetation regions.

Acknowledgements. This research was funded by

the China National Key Foundational Research and

Development Plan Program (2002CB111507) and the

China National Scientific and Technical Supporting Project

(2006BAD03A0206).

LITeRATURe CITeD

Bastin, C.N., G. Pickup, and G. Pearce. 1995. Utility of AVHRR

data for land degradation assessment: a case study. Int. J.

Remote Sens. 16: 651-672.

Birky, A.K. 2001. NDVI and a simply model of deciduous forest

seasonal dynamics. Ecol. Model. 143: 43-58.

Boyle, C.A., L. Lavkuli ch, H. S hre ier, and E. Kis s. 1997.

Changes in land cover and subsequence effects on lower

F res er Bas in e cosys tem from 1827 to 1990. Environ.

Manage. 21: 185-196.

Chen, J.M. and H.E. Wan. 2002. Vegetation Building and Water

& Soil Conservation of the Loess Plateau in China. Beijing:

Chinese forest Press, pp. 87-90.

Fang, J.Y. 2001. Re-dis cussion about the fores t vege tation

zonation in eastern China. Acta Bot. Sin. 43: 522-533.

Fang, J.Y. and K. Yoda. 1990. Climate and vegetation in China.

III. Water balance and distribution of vegetation. Ecol. Res.

5: 9-23.

Fu, C.B. and G. Wen. 1999. Variation of ecosystems over East

Asia in association with seasonal interannual and decadal

monsoon climate variability. Climate Change 43: 477-494.

Fuller, D.O. 1998. Foliar phenology of s avanna vegetation in

s outh-center Africa and its relevance related to climate

change, PhD Dissertation, University of Maryland, College

Park, Maryland.

Holdrige, L.R. 1947. Determination of world plant formation

from simple climate data. Science 105: 367-368.

Holdrige, L.R. 1976. Life Zone Ecology. San Jose, Costa Rica:

Tropical Science Center.

Hou, X.Y. 1983. Vegetation of China with re fe re nc e to its

geographical distribution. Ann. Missouri Bot. Gard. 70:

509-549.

Kimura, R., Y. Liu, N. Takayama, X. Zhang, M. Kamichika, and

N. Matsuoka. 2005. Heat and water balance the bare soil

surface and the potential distribution of vegetation in the

Loess Plateau, China. J. Arid Environ. 63: 439-457.

Li, D.Q., C.Y. S un, and X.S. Zhang. 1998. Dis tribution and

modeling of potential vegetation productivity in China. Acta

Bot. Sin. 40: 560-566.

Liu, H., T. Liu, and Z. Guo. 2003. Natural Vegetation of

Ge ographic al and Historical pe ri ods in the The L oess

Plateau. Chinese Sci. Bull. 48: 411-416.

Liu, T.S ., Z.T. Guo, N.Q. Wu, and H.Y. Lu. 1996. Prehistoric

vegetation on the The Loess P lateau: steppe or forest. J.

South. Asian Earth Sci. 13: 341-346.

Neilson, R.P. 1993. Vegetation redistribution: a possible

biosphere source of CO

2

during climatic change. Water Air

Soil Pollute 70: 659-673.

Neilson, R.P. 1995. A model for predicting continental- scale

vegetation distribution and water balance. Ecol. Appl. 5:

362-385.

Neilson, R.P. and D. Marks. 1994. A global perspective of

regional vege tation and hydrologic s ens itiviti es from

climatic change. J. Veg. Sci. 5: 715-730.

Prentice, I.C., T.S. Martin, and W. Cramer. 1993. A simulation

model for the transient effects of climate change on forest

landscapes. Ecol. Model. 65: 51-70.

P rentice, I.C., W. Cramer, S .P. Harrison, R. Leemans, R.A.

Monserud, and A.M. Solomon. 1992. A global biome model

based on plant physiology and dominance, soil properties

and climate. J. Biogeogr. 19: 117-134.

Ren, G.Y. 2004. On baseline vegetation in northern China. Acta

Ecol. Sin. 24: 1287-1293.

Rey, J.M. 1999. Modelling potent ial evapotrans piration of

potential vegetation. Ecol. Model. 12: 3141-3159.

Richard, Y. and I. Poccard. 1998. A statistical study of NDVI

sensitivity to seasonal and interannual rainfall variations in

southern Africa. Int. J. Remote Sens. 19: 2907-2920.

Running, S.W. and J .C. Coughlan. 1988. A ge neral model

of fores t ecos ys tem process for regional application. I.

Hydrologic balance. Canopy gas exchange and primary

production processes. Ecol. Model. 42: 125-154.

Schmidt, H. and A. Karnieli. 2000. Remote sensing of seasonal

variability of vegetation in a semi-arid environment. J. Arid

Eviron. 45: 43-59.

Shi, N. 1991. On distribution and change of historical forest of

China. Chinese Historical Geography 3: 43-73.

SUO et al. ¡X Vegetation deficiency in a typical region of the Loess Plateau in China

65

Sun, X.J. and Y.S. Chen. 1991. Palynological records of

Holocene Megathemal in China. Quaternary Science Rev.

10: 537-544.

Tans er, F.C. and A.R. P al mer. 1 999. T he appli cat ion of a

remotely sensed divers ity index to monitor degradation

pat tern in a se mi-a rid, hete rogeneo us, S outh Afric an

landscape. J. Arid Environ. 43: 477-484.

Tong, C., S. Yong, and W. Yong. 1996. Remote sensing analysis

of the ac cum ul at ed s no w dis a st ers i n t he t em pe rat e

range land. Act a S cie ntiarum Na tural ium Unie rs it atis

Neimonggo 27: 532-537.

Vegetation Atlas of China. 2001. Compiled by editorial board,

Chinese Academy of Sciences; Editors in chief: X.-Y. Hou

Beijing: Science Press.

Wang, Y.F. and X.M. Xiao. 1993. Climatic gradiant of main

vegetation types in the The Loess Plateau region. Acta Bot.

Sin. 35: 291-299.

Woodward, F.I. 1987. Climate and Plant Distribution. Cambridge

University Press, London.

Woodward, F.Y. and I.F. Mckee. 1991. Vegetaion and climate.

Environ. Int. 17: 535-546.

Wu, J. and R. Hobbs. 2002. Key issues and research priorities in

landscape ecology: an idiosyncratic synthesis. Landscape

Ecol. 17: 355-365.

Wu, Z.Y. 1980. China Vegetation. Beijing: Chinese Science

Press, pp. 749-1037.

Yu, G., X. Chen, J. Ni, R. Cheddadi, J. Guiot, H. Han, S .P.

Harrison, C. Huang, M. Ke, Z. Kong, S. Li, W. Li, P. Liew,

G. Liu, J. Liu, Q. Liu, K.-B. Liu, I.C. Prentice, W. Qui, G.

Ren, C. Song, S. Sugita, X. Sun, L. Tang, E. Van Campo, Y.

Xia, Q. Xu, S. Yan, X. Yang, J. Zhao, and Z. Zheng. 2000.

Palaeovegetation of China: a pollen data-based synthesis for

the mid-Holocene and last glacial maximum. J. Biogeogr.

27: 635-664.

Zeng, N., R.E. Dickinson, and X.B. Zeng. 1996. Climate impact

of Amazon deforestation¡Xa mechanis tic model study. J.

Clim. 9: 859-883.

Zhang, L.S. 1992. Formation, evolution and regional difference

of China¡¦s physic al geography. In M. Ren and H. Bao

(eds.), Physical regions of China and the exploitation and

m anagement Beijing: China S cience P ress , Vol. 25, pp.

36-69.

Zhang, X.S. 1993. A classification system of vegetation-climate

for global change research. Quaternary Sci. 28: 157-167.

Zhao, S.Q., H.N. Sun, R.J. Huang et al. 1988. Modern physical

geography. Beijing: Chinese Science Press, pp. 156-169.

Zhou, J.J. and L.H. Zhou. 1996. Analysis of vegetaion- climate

relations in Chaidamu region by Thornthwaite and Holdrige

potential evaporation index. Arid Region Res. 13: 46-51.

Zhao, C.Y., Z.R. Na n, a nd Z.D. F eng. 2004. GIS -ass is ted

s p a t i a l l y d i s t r i b u t e d m o de l i n g o f t h e p o t e n t i a l

evapotrans piration in s emi-arid clima te of the Chinese

Loess Plateau. J. Arid Environ. 58: 387-403.

Zhu, Z.C. 1983. Distribution of forest and stepped on the The

L oes s P lat eau. Acta Phytoe col. Geographic a Si n. 35:

122-131.

66

Botanical Studies, Vol. 49, 2008