Available via license: CC BY 4.0
Content may be subject to copyright.
Contents lists available at ScienceDirect
Journal of Rural Studies
journal homepage: www.elsevier.com/locate/jrurstud
Do settlements isolation and land use changes affect poverty? Evidence from
a mountainous province of China
Jianbin Xu
a,∗,1
, Jie Song
a
, Baochao Li
a
, Dan Liu
a
, Dong Wei
a
, Xiaoshu Cao
b,c
a
School of Geography and Planning, Sun Yat-sen University, Guangzhou, 510275, China
b
Shaanxi Normal University Academy of Natural Resources and Territorial Space, Shaanxi Normal University, Xi'an, 710119, China
c
Yunnan-Guizhou Plateau Observation Station of Coupled Human and Natural System, Shaanxi Normal University, Xi'an, 710119, China
ARTICLE INFO
Keywords:
Settlements isolation
Land use change
Poverty incidence
GWR
Guizhou
ABSTRACT
The impact of settlements isolation and land-use changes on poverty is often overlooked. This study explored the
spatial pattern and impact of settlements isolation and land use changes on poverty in Guizhou, a mountainous
province of China, based on a Pearson correlation analysis and geographically weighted regression (GWR). The
results showed that the incidence of high poverty levels in the counties of Guizhou tended to move from the
southwest to the southeast over the period of 2005–2015. Both settlements isolation and land use changes had an
impact on poverty, and this effect displayed spatial heterogeneity. With other factors unchanged, a 1 m increase
in settlements isolation was associated with a statistically significant 0.58% increase in poverty incidence.
Changes in the area of crop land, water, and forest land had a significant impact on poverty. The authors suggest
that in the process of resettlement, the distance between settlements should be kept less than 5000 m if possible.
The government needs to further improve and implement preferential land use policies, with appropriate in-
creases in the area of built-up land.
1. Introduction
Poverty remains a problem worldwide (Liu and Li, 2017b). Poverty
eradication is the first of the sustainable development goals (SDGs)
proposed by the United Nations. China has made a significant con-
tribution to the reduction of global poverty by lifting more than one-
fifth of the world's poor people out of poverty. Although the poverty
headcount ratio of individuals working for $1.90 a day or less
(2011 PPP) in rural China decreased from 88.3% in 1981 to 1.7% in
2018, China still has 16.6 million rural poor people (National Bureau of
StatisticsNBS, 2019). China has implemented several poverty allevia-
tion policies in recent years (Zhou et al., 2019), and has now entered a
critical stage of poverty eradication.
Most of the poverty-stricken counties in China are located in
mountainous areas (Liu and Xu, 2016). Around 70% of the poverty-
stricken counties in China are characterized by severe topographic
conditions, with an average slope of over 10° (Zhou and Xiong, 2018).
Recent research has focused on mountainous regions of China (Guo
et al., 2012;Gentle and Maraseni, 2012;Han et al., 2019). A remote
geographical location is usually considered to be the main cause of a
high poverty incidence in many mountainous areas (Bird and Shepherd,
2003;Zhou and Liu, 2019). Several studies have used distance from
major settlements and road network density as measures of settlements
remoteness (Okwi et al., 2007;Porterfield and McBride, 2007). The
spatial distribution of settlements affects residents’ability to enjoy in-
frastructure and services (Li and Wang, 2014), but isolation between
settlements is often ignored by researchers.
Natural topography determines land availability and regional ac-
cessibility and further influences the objective environment of wealth
creation in mountain areas (Liu et al., 2017a). Poverty-stricken counties
in mountainous areas depend on land for their wealth creation (Zhou
and Xiong, 2018). Land use is an important factor in rural social and
economic development (Liu et al., 2017b). With the rapid urbanization
experienced in China, many areas of crop land in peri-urban districts
have been transformed into urban built-up land (Liu et al., 2010). These
transformations have resulted in the disappearance of fertile land and
https://doi.org/10.1016/j.jrurstud.2020.04.018
Received 27 September 2019; Received in revised form 11 April 2020; Accepted 14 April 2020
∗
Corresponding author. Shaanxi Normal University Academy of Natural Resources and Territorial Space, Yunnan-Guizhou Plateau Observation Station of Coupled
Human and Natural System, Shaanxi Normal University, No. 620 West Chang'an Street, Chang'an District, Xi'an, Shaanxi, 710119, China.
E-mail addresses: xujb23@mail2.sysu.edu.cn (J. Xu), songj36@mail2.sysu.edu.cn (J. Song), libch9@mail2.sysu.edu.cn (B. Li), liud69@mail2.sysu.edu.cn (D. Liu),
weid5@mail2.sysu.edu.cn (D. Wei), caoxsh@mail.sysu.edu.cn (X. Cao).
1
Present/permanent address. School of Geography and Planning, Sun Yat-sen University No.135, Xingang West road, Gangzhou 510,275, Gangdong Province, P.R.
China.
Journal of Rural Studies xxx (xxxx) xxx–xxx
0743-0167/ © 2020 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/BY/4.0/).
Please cite this article as: Jianbin Xu, et al., Journal of Rural Studies, https://doi.org/10.1016/j.jrurstud.2020.04.018
reduced the sources of income available to farmers (Deng et al., 2015;
He et al., 2009). Another transformation trend is the conversion of
cultivated land to natural landscapes. For example, the Grain-for-Green
project provides incentives to farmers to convert cropland into forest
and grassland (Delang and Wang, 2013). There are also some projects
that have attempted to address ecological migration due to poor living
conditions, in which the government has assisted poverty-stricken
households to migrate to those towns or villages with better environ-
mental conditions, especially those with a central location (e.g., the
poverty alleviation resettlement (PAR) program in China) (Lo et al.,
2016). These projects have an impact on regional poverty. Land use
changes reflect the effectiveness of these projects, but the impact of
land use changes on poverty has not been well covered in the existing
research (Castella et al., 2005,2006;Liu et al., 2019).
Settlements are important social units in mountainous rural areas
because they reflect the relationships between people and the land, the
historical background, and the sociopolitical relationships (Yang et al.,
2016). The spatial distribution characteristics of settlements differ no-
tably, especially in poverty-stricken mountainous areas (Chen and Ge,
2015).
Few studies have examined the impact of settlements isolation and
land use changes on poverty, especially in mountainous areas. This
knowledge gap has prevented a full understanding of the spatial aspect
of poverty alleviation. To better understand the impact of settlements
isolation and land use changes in mountainous areas, and thus provide
a scientific reference for poverty reduction strategies and achieving the
SDGs in such areas, this study investigated the spatial pattern of set-
tlements isolation and land use changes in Guizhou, and the impact of
settlements isolation and land use changes on poverty.
The remainder of this paper is structured as follows: Section 2de-
scribes the study area and data collection methods. Section 3illustrates
the spatial patterns of poverty incidence, settlements isolation and the
changes of land use. The results are presented in Section 4. In Section 5,
the key implications of the study are addressed. Finally, conclusions are
presented in Section 6.
2. Study area, methods, and data
2.1. Study area
This study was conducted in Guizhou Province, in the eastern part
of the Yunnan-Guizhou Plateau (Fig. 1). The terrain tends to be high in
the west and low in the east, with an average elevation of around
1100 m. Guizhou Province has a subtropical plateau and humid mon-
soon climate. Mountainous and hilly areas account for 90% of the total
area of the province, of which a karst landform area accounts for 70%
(Wang et al., 2017). Because of the mountainous natural environment,
the size of the rural poor population that needs to be relocated in
Guizhou Province is the highest of all provinces in China according to
the 13
th
Five-Year Plan for the Ex Situ Poverty Alleviation Relocation of the
National Development and Reform Commission of China (National
Development and Reform Commission of China (NDRC), 2016). Ac-
cording to the 2016 Statistical Yearbook of Guizhou Province, the poverty
incidence in Guizhou Province was 14% in 2015. Of all the counties in
the province, 10.4% were classed as poverty-stricken by the Chinese
State Council Group Office of Poverty Alleviation and Development and
were therefore considered to be in need of poverty alleviation and
development. In 2015, there were 4.93 million poor people residents in
the province, accounting for 8.77% of the whole country's poor.
Guizhou Province has the largest number of rural poor in China. After
2020, China will eradicate absolute poverty, but relative poverty will
persist for a long time. It is necessary to implement more targeted
poverty reduction measures from a spatial perspective, especially with
regard to settlements and land use.
2.2. Methods and data
It was necessary to test the correlation of factors before performing
a regression analysis, and we therefore used Pearson's correlation
coefficient to test the correlation between settlements isolation, land
use changes, and poverty. By referring to related research (Castella
et al., 2005), to better condense the information and conduct a cross-
variation analysis of variables, we used a principal component analysis
(PCA) to analyze the internal relationship between settlements isola-
tion, land use changes, and poverty. Finally, the spatial heterogeneity of
poverty influencing factors, especially settlements isolation and land
use changes, was analyzed by a geographically weighted regression
model (GWR).
2.2.1. Estimation of settlements accessibility
With reference to existing research, we added indicators of settle-
ments isolation in addition to Road Density and Remoteness (Castella
et al., 2005). To measure the isolation of settlements, we obtained
settlements data and administrative boundary data. The county was the
unit used in this research. Vector map data was obtained from the
National Catalogue Service For Geographic Information (http://www.
webmap.cn/main.do?method=index). This is a 1:1 million map that
provides settlements data, including ordinary houses, yurts, and
herding points. The kernel density analysis method in ArcGIS Desktop
was used to interpolate the road network in Guizhou Province, and the
average line density within each county is extracted as the Road Density
variable. Isolation was determined as the cumulative equity index as
measured by travel distance to the closest settlement, calculated by the
average value in the county. Remoteness was determined the average
travel distance to the district administrative center to which the set-
tlement belongs and to Guiyang City, which is the capital city of
Guizhou.
2.2.2. Land use and poverty data
Land use data with a 500 m resolution was obtained from the
Geographical Information Monitoring Cloud Platform (http://www.dsac.
cn/DataProduct/Detail/200804). The average classification accuracy of
crop, urban and rural, industrial, mining, and residential land was more
than 85%, and the average classification accuracy of other land use
types was more than 75%, which were suitable for the purposes of this
study. Using the Guizhou land use data, the proportion of each land use
class of each county was calculated for 2005 and 2015, resulting in 12
variables (Crop_15, Forest_15, Grass_15, Water_15, Construction_15,
Unutilized_15, etc.). The changes in land use between each pair of dates
was calculated to obtain six variables (Crop 05_15, Forest 05_15, Grass
05_15, Water 05_15, Construction 05_15, Unutilized 05_15). Using the
ArcGIS grid computing tool, a land use conversion matrix for the period
of 2005–2015 was calculated, and land use change variables were ob-
tained for each county.
Poverty incidence data for each county in 2005 and 2015 was ob-
tained from the 2006 and 2016 Guizhou Statistical Yearbook (Poverty_05
& Poverty_15). The poverty incidence indicates the proportion of the
poor population in a given area, which refers to the proportion of the
total population whose per capita income or consumption expenditure
does not meet the accepted poverty standard (Xian et al., 2016).
2.2.3. Geographically weighted regression
A GWR was applied to estimate the impact on different regions and
reflect the spatial non-stationarity of parameters in different spaces,
enabling the results to be more realistic. The model was as follows:
∑
=+ +yβuv βuvx ε(, ) (, )
iii kkiiik
i
0(1)
where y
i
is the observation,;
β
uv(,
)
ii
0
is the regression coefficient for
point i, which indicates the degree of influence of independent variables
on y
i
;
uv
(
,
)
ii
is the geographic center coordinate of the isample space
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
2
unit;
β
uv(,
)
k
is the value of the continuous function
β
uv(,
)
ii
0
in lo-
cation i;x
ik
represents the value of an independent variable at point i;
ε
is a normal distribution function with constant variance, representing a
random error term. Because the Akaike information criterion (AIC)
method considers the differences in the degrees of freedom between
different models, the Gaussian function was used to determine the
weight and the AIC method was used to determine the optimal band-
width (Yang et al., 2015).
3. Spatial patterns of poverty incidence, settlements isolation, and
changes of land use
3.1. Spatial pattern of poverty incidence
The poverty incidence in Guizhou Province in 2005 and 2015 was
visualized in the ArcGIS desktop software (Fig. 2).
An assessment of the spatial distribution of poverty incidence in
Guizhou Province revealed that the poverty incidence was highest in
the eastern, western, and southern counties, and lowest in the central
and northern counties (Fig. 2). Some researchers have described the
spatial heterogeneity of poverty in Guizhou Province as having a “horse
hoof”shaped distribution (Wang et al., 2017). In 2005, the areas with a
high poverty incidence were mainly distributed in Liupanshui City
(Shuicheng County), southwestern Buyi, Miao Autonomous Prefecture,
and Bijie City (Nayong County, Zhijin County) (Fig. 2a). These counties
are located in remote mountain areas, such as the Wumeng Mountains
in the northwest and Miaoling in the south. In 2015, the poorest
counties were in areas such as the Southeast Guizhou Miao and Dong
Autonomous Prefecture, which are located in southeast Guizhou
(Congjiang County, Rongjiang County) (Fig. 2b). The high poverty in-
cidence in these areas was because these counties are located in
Miaoling, where there is large population of ethnic minorities and the
topographic conditions limit people's livelihoods. As can be seen from
the violin chart in Fig. 3a, the distribution of poverty incidence was
more concentrated in 2005, with most counties having a poverty in-
cidence of around 10%. After the implementation of the 10-year pov-
erty alleviation policy, the distribution of poverty incidence in 2015
was more dispersed, although there was a concentration of counties in
the 10–25% range. Fig. 3b shows that there was no consistency in the
changes in the ranking of poverty incidence among the Guizhou
counties. The poverty ranking of the poorest counties in 2005 decreased
in 2015, with the ranking of counties with poverty incidence rates
ranked between 66 and 77 in 2005 increasing to above 77 in 2015. This
might because the government implemented various poverty reduction
measures in the poorest counties in 2005–2015, such as developmental
poverty alleviation being implemented for the working poor, poverty
alleviation in areas with sub-standard living conditions being im-
plemented by ex-situ relocation poverty alleviation, and labor training
and the transfer of poverty alleviation. Poor people that were unable to
find work were helped through the establishment of an employment
assistance system.
3.2. Spatial pattern of settlements isolation
The kernel density analysis of the settlements revealed that the
settlements in the central and northern counties of Guizhou Province
were densely distributed (Fig. 4a). The density of settlements in the
south, southeast, and southwest of Guizhou was lower than in the
north, which was related to the large population of ethnic minorities
and the mountainous terrain. From the perspective of settlement ac-
cessibility in the counties, the highest density road network was con-
centrated in Guiyang, the capital city of Guizhou Province. The road
Fig. 1. The study area.
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
3
network density in western and northwestern Guizhou was lower than
that in the central area; however, it was lower in southeastern Guizhou,
which had a high poverty incidence (Fig. 4b). The highest settlements
isolation was concentrated in Huishui County, Pingtang County,
Dushan County, Sandu Shui Autonomous County and other counties in
southern Guizhou, as well as Chishui City, Xishui County, Tonglu
County, Shuyang County, Zheng'an County, and Daozhen County in the
north of Guizhou, most which are located in mountainous areas such as
the Wumeng Mountains in western Guizhou and Miaoling in southern
Guizhou. The counties with the lowest settlements' isolation were
mainly concentrated in eastern and western Guizhou (Fig. 4c), which
may be due to the smaller territory of these counties and the relatively
shorter distance between settlements. There was a symmetrical law
explaining the distribution of remoteness among county settlements,
with the axis of symmetry being northwest-central-southeast of
Guizhou, and the degree of remoteness from both sides of this axis
gradually increasing (Fig. 4d).
3.3. Land use changes
According to the Guizhou land use raster data for 2005 and 2015, a
land use transfer matrix (Table 1) and the land use structure (Fig. 5)
from 2005 to 2015 was extracted.
The land use transfer matrix (Table 1) shows that the change in the
area of unutilized land was largest in 2005–2015, and that the area
converted to built-up land was 2 km
2
(5.41%). This was mainly because
since the mid-1980s, Guizhou Province has implemented a series of
ecological restoration measures, such as soil and water conservation,
natural restoration, and the conversion of sloping land to small wa-
tersheds. The area converted to grassland and forest land was 101 and
99 km
2
, respectively, and the area converted to water bodies was
35 km
2
(Table 1). The trend of rock desertification in Guizhou Province
has been restricted to some extent. According to Zhang (2016),asof
2014, the cumulative investment in the natural forest resource pro-
tection project implemented in Guizhou Province was 3.147 billion
yuan, and the cumulative investment in the returning farmland to forest
project was 18.08 billion yuan. A rock desertification comprehensive
treatment project commenced in 2008, with an investment of 1.35
billion yuan as of 2010 (Zhang, 2016). Under the influence of these
projects, the area converted from grassland to forest land was 338 km
2
.
The area converted into cultivated land and water bodies was 34 and
30 km
2
, respectively. The rate of change to grassland during this period
was 1.93%. The area of crop land converted into built-up land re-
presented the largest conversion, accounting for a total of 591 km
2
. This
was consistent with the trend observed in the existing research, which
shows that the urbanization process has consumed a large amount of
crop land (Long et al., 2009).
Rapid urbanization has led to a frenzied expansion of built-up land.
The urbanization rate of Guizhou Province increased from 26.87% in
2005 to 42.01% in 2015, an increase of 15.14%. This expansion oc-
curred through a large conversion of cultivated land to built-up land
(Fig. 5). The expansion of urbanization has caused a reduction in the
ecological function of land cover types such as crop land, forest land,
and grassland. Although the overall land output value has increased,
the impact on rural areas, agriculture, and farmers has also been ex-
tensive. As a consequence, there has been a diversification of farmers’
Fig. 2. The spatial pattern of poverty incidence. (2a. Poverty incidence in 2005 (%); 2 b. Poverty incidence in 2015 (%)).
Fig. 3. Distribution (a) and rank changes (b) in poverty incidence (The rank of the highest poverty incidence is 88 at the top of the vertical axis in Fig. 3b).
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
4
livelihood strategies, with urbanization attracting more farmers to work
in cities. According to the second agricultural census in Guizhou Pro-
vince, there were about 4.41 million migrant workers in the province in
2006 and 8 million in 2012. The loss of the young and middle-aged
labor force has led to a reshaping and transformation of the rural
economy and society. The “hollow village”and “empty nest elderly”
phenomena have become widespread, and vulnerable rural groups,
such as the old and children, have become significantly poorer.
4. Results
4.1. Impact of settlements isolation on poverty
There was a significant positive correlation between settlements
isolation (Isolation) and poverty incidence in 2005 and 2015 (Poverty_05
& Poverty_15)(Table 2). This result indicates that counties with a higher
poverty incidence in 2005 and 2015 were mainly located in areas with
high settlements isolation. The spatial distribution of settlements in
poverty-stricken areas had an important correlation with poverty.
There was a spatial spillover effect in the development of the regional
economy (Zhang et al., 2016). The spatial distance between residential
areas will have an effect of enhancing or weakening this type of space
spillover. Li and Wang (2014) suggested that the degree of dispersion of
settlements had a negative correlation with the net income of residents,
especially in mountainous areas. The greater the isolation, the more the
impact of developed settlements on poor settlements was blocked,
which resulted in poor areas becoming even poorer.
There was a significant negative correlation between remoteness
and road density. The areas with a high road network density were
mostly located in areas with low remoteness. These counties were
generally concentrated around Guiyang, which is the capital city of
Guizhou Province, where the transportation network is more accessible,
and the degree of remoteness is lower.
Road density (Road density) had a negative correlation with poverty
incidence in 2005 and 2015 (Poverty_05 & Poverty_15). This correlation
illustrates that even with the rapid construction of transportation
Fig. 4. AKernel density analysis of settlements and the spatial pattern of accessibility (The results of an analysis of the kernel density of settlements (Fig. 4a), road
network density (Fig. 4b), isolation of settlements (Fig. 4c), and remoteness of settlements (Fig. 4d) for each county.).
Table 1
Land use transfer matrix 2005–2015 (km
2
).
2005/2015 Crop land Forest land Grass land Water bodies Built-up land Unutilized land Rate/%
Crop land 48588.00 99.00 101.00 35.00 591.00 2.00 1.68
Forest land 80.00 94169.00 298.00 28.00 241.00 0.00 0.68
Grass land 30.00 338.00 30220.00 34.00 193.00 0.00 1.93
Water bodies 0.00 0.00 0.00 398.00 1.00 0.00 0.25
Built-up land 0.00 1.00 0.00 1.00 577.00 0.00 0.35
Unutilized land 0.00 0.00 0.00 0.00 2.00 35.00 5.41
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
5
networks, transport restrictions still have a certain explanatory power
for poverty incidence. The road network is a support system for regional
economic development. In addition, the limited financial capacity of
poverty-stricken areas will result in an insufficient infrastructure con-
struction capacity, which in turn will limit market accessibility and lead
to poor areas becoming poorer.
4.2. Impact of land use changes on poverty
The analysis of correlations between land use changes and poverty
revealed that the change of crop land area (Crop 05_15) and built-up
land during 2005–2015 (Construction 05_15) was positively correlated
(Table 2). This means that where there was a large change in crop land
area, there was also a large change in built-up land, and vice versa.
Changes in the area of crop land were mainly concentrated in the vi-
cinity of urban built-up areas, indicating that there were constraints on
the grain production space due to urbanization in Guizhou Province.
The changes in the area of crop land during 2005–2015 (Crop 05_15)
had a significant negative correlation with poverty incidence in 2005
and 2015 (Poverty_05 & Poverty_15). The counties with a high poverty
incidence had a low change in crop land area, while the counties with a
high poverty reduction rate had a large change in the area of crop land
(Castella et al., 2005). The rate of change in the area of crop land in
poor counties was small, indicating that agriculture was still the main
industry in poor areas.
The PCA results shown in Fig. 6 revealed that poverty incidence
(Poverty_05 & Poverty_15) was positively correlated with the remoteness
and isolation of settlements, and negatively correlated with variables
such as the road network density (Road density) and built-up land
(Construction _15). This principal component indicated that the longer
the distance between settlements or the farther away they were from
the administrative center, the more prone they were to poverty. Fig. 7
also reveals that the higher the proportion of crop land, forest land,
grassland, and utilized land in the county, the poorer the county was,
because the poverty variables and the arable land (Crop _15), forest land
(Forest _15), grassland (Grass _15), and other variables were located in
the positive axis half of the first principal component.
In the first principal component axis (horizontal axis), the variable
of grassland (Grass 05_15) was also positively correlated with the
poverty variables. This shows that grassland changes mainly occurred
in poorer areas during 2005–2015. The direction of grassland transfer is
shown in Table 1, with grassland mainly converted into forest land,
built-up land, and cultivated land. The rapid urbanization has led to a
rapid expansion of built-up land, which has consumed large areas of
forest land and cultivated land. On the other hand, due to the shortage
Fig. 5. Chord diagram representing the land use conversion during 2005–2015
(To show the transition between different types of land use, the figure shows
the logarithmic value of different types of land use area.).
Table 2
Pearson correlation matrix.
Isolation Remoteness Road density Crop 05_15 Grass 05_15 Unutilized 05_15 Crop
_15
Construction
_15
Forest _15 Water
_15
Grass
_15
Remoteness 0.521**
Road density −0.747** −0.591**
Crop 05_15 −0.257* 0.245*
Crop_15 0.385** 0.360** −0.402** 0.276** 0.253*
Water 05_15 0.470**
Construction 05_15 −0.637** −0.418** 0.629** 0.365** 0.316**
Forest 05_15 0.685** 0.511** −0.651** 0.644** −0.370**
Grass_15 0.243** 0.225* 0.497** 0.376**
Unutilized_15 0.255** 0.339** 0.253*
Poverty_15 0.551** −0.575** −0.326** −0.497** 0.366** −0.217* 0.261*
Poverty_05 0.513** −0.485** −0.357** −0.529** 0.273* 0.333**
**p < 0.01 * p < 0.05.
Fig. 6. Principal component analysis (PCA) of accessibility, land use, and
poverty (The first two axes account for 42% of the total variance, with axis 1
accounting for 28% and axis 2 13%.).
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
6
of built-up land in the more developed areas, through the increasing vs.
decreasing balance land-use policy between urban and rural built-up
land and ecological compensation measures, there has been an impact
on forest land and grassland in remote and poor areas.
Combining Table 1 and Fig. 6, it was apparent that poverty in-
cidence was negatively correlated with the change of crop land vari-
ables (Crop 05_15). The transfer direction of crop land during this
period was mainly a conversion into built-up land, followed by grass-
land and forest land. This trend indicated that the greater the rate of
change of crop land, the lower the poverty incidence.
4.3. Spatial heterogeneity analysis of the impact of settlements isolation and
land use change on poverty
According to the correlation coefficient matrix presented in Table 2,
Isolation,Road density,Crop 05_15,Water_15, Construction_15, and
Forest_15 were selected as independent variables, poverty incidence
(Poverty_15) was used as the dependent variable, and the spatial het-
erogeneity of the influence of the independent variable on the depen-
dent variable was analyzed by the GWR model. Prior to the GWR re-
gression analysis, a global ordinary least squares (OLS) test was
performed. In the OLS analysis results (Table 3), the multiple R-squared
and adjusted R-squared values explained 37% and 33% of the variation
Fig. 7. Geographically weighted regression (GWR) coefficients.
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
7
of the dependent variable, respectively. At the same time, the variance
inflation factor (VIF) values of all the independent variables were less
than 7.5, which indicated that there were no redundant variables in the
model. The results of a Jarque-Bera test were not statistically sig-
nificant, indicating that model predictions were comprehensive.
A local spatial regression was performed using GWR. The small-
sample corrected AIC (AICc) value of the GWR was 211.34, which was
significantly smaller than the OLS regression result (if the difference in
the AIC value between two models is larger than 3, the model with the
lower AICc value is considered a better model). The multiple R-squared
value was 0.59 and the adjusted R-squared value was 0.47. Both GWR
parameters produced better results than the OLS (Table 4).
The GWR revealed a global positive correlation between settlements
isolation and poverty incidence, except for Daozhen County (Fig. 7a). A
1 m increase in the distance between settlements was associated with a
statistically significant 0.58% increase in poverty incidence. This con-
clusion can provide a quantitative reference for poverty alleviation
resettlement and settlements consolidation (Lo et al., 2016). The areas
with a positive correlation between settlements isolation and poverty
incidence were mainly concentrated in southeast Guizhou and south-
west Guizhou. These two regions are minority autonomous prefectures.
Because minority settlements are mostly located in mountainous areas,
they are more dispersed and are not conducive to agricultural activities,
which results in high incidence of poverty.
The Road_density was positively correlated with Poverty_15 in more
than 89% of counties, with these areas mainly concentrated in less
impoverished counties (Fig. 7b). In poverty-stricken counties in eastern
and southern Guizhou, road density and poverty incidence were ne-
gatively correlated. A high-density road network indicates a flatter
natural terrain, and also reflects a high level of infrastructure con-
struction in the county. This will attract external investment and pro-
mote economic and social development in the region, thereby reducing
the poverty incidence.
The estimated distribution of the parameters of Crop15_15 showed a
global negative correlation (Fig. 7c). This trend indicates that the
poverty incidence was lower in areas with larger changes in crop land.
Combining Fig. 6 and Table 1, it was apparent that from 2005 to 2015,
crop land was mainly converted into built-up land, grassland, and forest
land. This reflects the poverty reduction generated by urbanization, but
also illustrates the role of the “Grain-for-Green”and soil and water
conservation projects in poverty reduction.
The area of water bodies and poverty incidence were negatively
correlated in 88% of counties (Fig. 7d). Due to the wide distribution of
karst landforms in Guizhou, the province is neither drought-tolerant nor
water-tolerant. Drought has a greater impact on rice yield during the
heading and flowering stages. Waterlogging has a direct impact on the
livelihood and property of residents (Guo and Jiang, 1995;Hu et al.,
2019). Therefore, the abundance of water resources also has an im-
portant impact on spatial poverty in Guizhou Province.
The proportion of built-up land in 96% of counties was negatively
correlated with poverty incidence, indicating the significance of in-
dustrialization and urbanization in poverty reduction (Fig. 7e). The
area of built-up land has rapidly expanded during the 21st century in
Guizhou. With the rapid urbanization process, a large number of rural
laborers from poor areas have migrated to cities to increase their in-
come levels (Liu et al., 2015).
The proportion of forest area and poverty incidence were negatively
correlated in 86% of counties, which may a consequence of the ecolo-
gical poverty alleviation policy in central and western China (Fig. 7f).
Local people can serve as forest rangers, poor families can grow cash
crops, and regional ecological compensation measures have accelerated
the pace of poverty alleviation in areas with a high poverty incidence.
For example, the central government provides 150 kg of food and 20
yuan in cash to subsidize every mu (1 ha = 15 mu) of land converted
(Xu et al., 2007). The land conversion project, which receives continued
financial support, also contributes to the development of the rural
economy and therefore promotes poverty reduction (Wu et al., 2019).
5. Discussion
5.1. General discussion
Previous studies have focused on the connection between settle-
ments and the locational factors, such as the accessibility of settlements
and their road density. Poverty reduction measures that consider exo-
genous factors are appropriate, but the findings of this study proved
that the issue of isolation between settlements requires more attention.
There are also various factors that can cause land use changes, espe-
cially the conversion of crop land to other uses, and these changes are
closely related to poverty incidence. It is therefore necessary to explore
these factors and their role in reducing poverty.
5.1.1. Settlements isolation in poverty-stricken counties needs more
attention
Table 5 shows the results of a settlements isolation analysis in
poverty-stricken and non-poverty-stricken counties. The data for pov-
erty-stricken counties were obtained from the 2016 Guizhou Statistical
Yearbook. The average settlements isolation of poverty-stricken coun-
ties reached 7248 m, but the settlements isolation of non-poverty-
stricken counties was only 5656 m. This means that if settlements iso-
lation was less than 5656 m, the county was more likely to be non-
poverty stricken. Fig. 8a shows the linear relationship between the
isolation of settlements and poverty incidence in 2015, indicating that
the greater the isolation of settlements, the higher the poverty in-
cidence.
Fig. 8b shows the differences in the distribution of isolation among
different types of county, with the peak value of settlements isolation in
poverty-stricken counties occurring at about 7600 m, while in non-
poverty-stricken counties it occurred at about 6400 m. These results
indicated that most of the settlements in poverty-stricken counties were
more isolated than less-poor settlements in other counties. From the
Table 3
The ordinary least squares (OLS) test results.
R
2
Adjusted R
2
AICc VIF Jarque-Bera
0.37 0.33 523.23 < 7.5 p > 0.01
Table 4
The geographically weighted regression (GWR) results.
Variable Min 25% quantile Median 75% quantile Max
Isolation −0.01 0.46 0.58 0.63 0.81
Road density −0.36 0.33 0.81 1.25 1.61
Crop 05_15 −0.93 −0.63 −0.36 −0.25 −0.03
Water_15 −0.50 −0.45 −0.39 −0.26 0.40
Construction_15 −0.76 −0.30 −0.20 −0.15 0.12
Forest_15 −1.10 −0.83 −0.50 −0.17 0.33
R
2
0.71
Adjusted R
2
0.59
AICc 211.34
Bandwidth 174239.17
Table 5
Descriptive analysis of settlements isolation in different types of counties.
Poverty-stricken-counties Non-poverty-stricken-counties
Counts/Percentage 50(57%) 38(43%)
Mean(m) 7248 5656
Min(m) 5108 2063
Max(m) 10,180 7983
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
8
perspective of reducing poverty, settlements with a high degree of
isolation should be relocated to ensure that better infrastructure and
services can be provided, enabling the residents to escape from poverty.
The necessity of resettlement for poverty alleviation in remote moun-
tain areas has been reported previously (Lo et al., 2016). Natural geo-
graphical conditions, as well as historical and cultural factors both lead
to settlements isolation. For example, ethnic minorities are more con-
centrated in mountain areas. In addition to living habits, the sense of
belonging to place is also an important factor in the formation and
development of settlements (Bocco, 2016). In addition, the isolation of
settlements also limits the ability to break the spatial poverty trap. The
successful experience of settlements that have moved out of poverty
will have the effect of providing a positive demonstration. A high level
of settlements isolation will cause chronic poverty.
5.1.2. Land use changes and poverty
Table 1 and Fig. 6 show that the main conversion of crop land was
to built-up land, grassland, and forest during 2005–2015. The area of
built-up land has reached 591 km
2
, accounting for 71% of the con-
version area, and the area of grassland and forest was about 200 km
2
,
accounting for 24% of the conversion area. The conversion of crop land
to built-up land has occurred through the rapid urbanization process,
while the conversion to forests and grasslands represents the im-
plementation of ecological protection policies. Urbanization has re-
sulted in a rapid expansion of urban land (Bai et al., 2014;Fang et al.,
2017). The increase in the area of built-up land has improved the in-
vestment environment, provided more opportunities for the non-agri-
cultural employment of the poor, and effectively promoted urban de-
velopment and economic growth (Li et al., 2020). At the same time, as
mentioned above, the conversion of crop land to forest land and
grassland is an important outcome of the implementation of the “Grain-
for-Green”policy. Such ecological protection policies have had a com-
prehensive effect on poverty reduction, not only protecting the en-
vironment of fragile ecological areas, but also directly increasing
farmers’income.
5.1.3. Implications for poverty-alleviation-oriented settlements and land use
policies
The impact of settlements isolation on poverty is very important. In
the resettlement process, attention needs to be paid to the degree of
isolation between settlements. Based on the experience of Guizhou, the
distance between settlements should be less than 5000 m if possible. In
the process of resettlement, in addition to considering natural factors
and the satisfaction of residents, the sense of belonging of residents
should also be considered. For settlements where such migrations
would prove difficult, improvements should be made to the infra-
structure, the isolation of settlements should be reduced, the
“Polarization and trickledown effect”should be strengthened, and the
spatial poverty trap should be broken. For poverty-stricken areas with a
high level of ecological sensitivity, especially in rocky desertification
areas, the return of cropland to forests and other ecological restoration
measures should be intensified by raising subsidy levels. Some re-
searchers have proposed the return of farmland sloping > 25° and the
rocky desertification mountainous area in Guizhou Province, a com-
prehensive assessment of the scale of the potential returned farmland
and the direction of industrial development, and an improvement in the
level of subsidy provided for returning farmland to forests (Zhang,
2016).
In terms of land use, the government should increase built-up land
quotas for poverty-stricken counties. This is because under rapid ur-
banization, a large amount of crop land can be converted to built-up
land. More effort should be made to increase the scale of built-up land
in urban land use planning and to increase the use of special land use
quotas for poverty alleviation (Li et al., 2020).
6. Conclusions
1) Settlements isolation was significantly positively correlated with
poverty incidence. With other factors unchanged, a 1 m increase in
settlements isolation was associated with a statistically significant
0.58% increase in poverty incidence.
2) Land is the main source of income for poor people in poverty-
stricken areas, and changes in land use will affect poverty incidence.
Changes in crop land have promoted poverty reduction, mainly due
to poverty reduction strategies such as the “Grain-for-Green”project
and urbanization in poverty-stricken areas. Urbanization and eco-
logical protection policies have promoted poverty reduction in
mountainous areas.
3) Due to the widespread prevalence of karst landforms in Guizhou
Province, water resources are a limiting factor in regional poverty
alleviation. Regions with larger water bodies had a lower poverty
incidence.
4) In the resettlement process, based on the experience of Guizhou, the
distance between settlements should be less than 5000 m if possible.
The government needs to further improve and implement pre-
ferential land use policies and appropriately increase the quotas of
built-up land.
Fig. 8. Isolation and poverty (Linear fitting of Poverty_15 and Isolation, the values are normalized (Fig. 8a), b. Density distribution of Isolation (Fig. 8b)).
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
9
CRediT authorship contribution statement
Jianbin Xu: Conceptualization, Methodology, Software, Writing -
review & editing. Jie Song: Data curation, Writing - original draft.
Baochao Li: Visualization, Investigation. Dan Liu: Software,
Validation. Dong Wei: Writing - review & editing. Xiaoshu Cao:
Supervision.
Acknowledgements
This work was supported by the National Natural Science
Foundation of China [No. 41831284].
References
Bai, X., Shi, P., Liu, Y., 2014. Society: realizing China's urban dream. Nature 509, 158.
https://doi.org/10.1038/509158a. 7499.
Bird, K., Shepherd, A., 2003. Livelihoods and chronic poverty in semi-arid Zimbabwe.
World Dev. 31, 591–610. https://doi.org/10.1016/S0305-750X(02)00220-6.
Bocco, G., 2016. Remoteness and remote places. a geographic perspective. Geoforum 77,
178–181. https://doi.org/10.1016/j.geoforum.2016.11.003.
Castella, J.C., Boissau, S., Thanh, N.H., Novosad, P., 2006. Impact of forestland allocation
on land use in a mountainous province of Vietnam. Land Use Pol. 23https://doi.org/
10.1016/j.landusepol.2004.07.004. 0-160.
Castella, J.C., Manh, P.H., Kam, S.P., Villano, L., Tronche, N.R., 2005. Analysis of village
accessibility and its impact on land use dynamics in a mountainous province of
northern Vietnam. Appl. Geogr. 25, 308–326. https://doi.org/10.1016/j.apgeog.
2005.07.003.
Chen, Y., Ge, Y., 2015. Spatial point pattern analysis on the villages in China's poverty-
stricken areas. Procedia Environ. Sci. 27, 98–105. https://doi.org/10.1016/j.proenv.
2015.07.098.
Delang, C.O., Wang, W., 2013. Chinese forest policy reforms after 1998: the case of the
natural forest protection program and the slope land conversion program. Int. For.
Rev. 15, 290–304. https://doi.org/10.1505/146554813807700128.
Deng, X., Huang, J., Rozelle, S., Zhang, J., Li, Z., 2015. Impact of urbanization on cul-
tivated land changes in China. Land Use Pol. 45, 1–7. https://doi.org/10.1016/j.
landusepol.2015.01.007.
Fang, C., Li, G., Zhang, Q., 2017. The variation characteristics and control measures of the
urban construction land in China. J. Nat. Resour. 32, 363–376. https://doi.org/10.
11849/zrzyxb.20160361. (in Chinese).
Gentle, P., Maraseni, T.N., 2012. Climate change, poverty and livelihoods: adaptation
practices by rural mountain communities in Nepal. Environ. Sci. Pol. 21, 24–34.
https://doi.org/10.1016/j.envsci.2012.03.007.
Guo, L., Jiang, D., 1995. A study on classification and development of the poor areas in
China. Geogr. Res. 14, 1–7. (in Chinese). http://en.cnki.com.cn/Article_en/
CJFDTotal-DLYJ502.000.htm.
Guo, X., Zhang, Q., Ma, L., 2012. Analysis of the spatial distribution character and its
influence factors of rural settlement in transition-region between mountain and hilly.
Econ. Geogr. 32, 114–120. (in Chinese). http://en.cnki.com.cn/Article_en/
CJFDTOTAL-JJDL201210018.htm.
Han, H., Gao, H., Huang, Y., Chen, X., Chen, M., Li, J., 2019. Effects of drought on
freshwater ecosystem services in poverty-stricken mountain areas. Glob. Ecol.
Conserv. 17, e00537. https://doi.org/10.1016/j.gecco.2019.e00537.
He, S., Liu, Y., Webster, C., Wu, F., 2009. Property rights redistribution, entitlement
failure and the impoverishment of landless farmers in China. Urban Stud. 46,
1925–1949. https://doi.org/10.1177/0042098009106015.
Hu, J., Gu, S., Zhang, B., Zhao, J., 2019. Rice drought loss assessment under karst
landform in Guizhou province. Guizhou Agricultural Sciences 47, 138–141. (in
Chinese). http://en.cnki.com.cn/Article_en/CJFDTotal-GATE201905030.htm.
Li, H., Wang, Y., 2014. Discrete degree on village settlement's spatial distribution and its
correlation with net income of rural residents in poverty county. Geogr. Res. 33,
1617–1628. http://en.cnki.com.cn/Article_en/CJFDTotal-DLYJ201409004.htm.
Li, Y., Li, Y., Karácsonyi, D., Liu, Z., Wang, Y., Wang, J., 2020. Spatio-temporal pattern
and driving forces of construction land change in a poverty-stricken county of China
and implications for poverty-alleviation-oriented land use policies. Land Use Pol. 91,
104267. https://doi.org/10.1016/j.landusepol.2019.104267.
Liu, T., Liu, H., Qi, Y., 2015. Construction land expansion and cultivated land protection
in urbanizing China: insights from national land surveys, 1996–2006. Habitat Int. 46,
13–22. https://doi.org/10.1016/j.habitatint.2014.10.019.
Liu, Y., Cao, X., Xu, J., Li, T., 2019. Influence of traffic accessibility on land use based on
Landsat imagery and internet map: a case study of the Pearl River Delta urban ag-
glomeration. PloS One 14https://doi.org/10.1371/journal.pone.0224136.
e0224136-e0224136.
Liu, Y., Li, Y., 2017. Revitalize the world's countryside. Nature 548, 275. https://doi.org/
10.1038/548275a.
Liu, Y., Liu, J., Zhou, Y., 2017. Spatio-temporal patterns of rural poverty in China and
targeted poverty alleviation strategies. J. Rural Stud. 52, 66–75. https://doi.org/10.
1016/j.jrurstud.2017.04.002.
Liu, Y., Liu, Y., Chen, Y., Long, H., 2010. The process and driving forces of rural hollowing
in China under rapid urbanization. J. Geogr. Sci. 20, 876–888. https://doi.org/10.
1007/s11442-010-0817-2.
Liu, Y., Xu, Y., 2016. A geographic identification of multidimensional poverty in rural
China under the framework of sustainable livelihoods analysis. Appl. Geogr. 73,
62–76. https://doi.org/10.1016/j.apgeog.2016.06.004.
Lo, K., Xue, L., Wang, M., 2016. Spatial restructuring through poverty alleviation reset-
tlement in rural China. J. Rural Stud. 47, 496–505. https://doi.org/10.1016/j.
jrurstud.2016.06.006.
Long, H., Zou, J., Liu, Y., 2009. Differentiation of rural development driven by in-
dustrialization and urbanization in eastern coastal China. Habitat Int. 33, 454–462.
https://doi.org/10.1016/j.habitatint.2009.03.003.
National Bureau of Statistics, NBS, 2019. National Bureau of Statistics: decrease 13.86
million people in rural poverty in 2018. Retrieved March 5, 2020, from. http://www.
xinhuanet.com/fortune/2019-02/15/c_1124120302.htm.
National Development and Reform Commission of China (NDRC), 2016. The 13th five-
year plan for the ex situ poverty alleviation relocation. Retrieved March 5, 2020,
from. http://www.cpad.gov.cn/art/2017/4/28/art_50_62482.html.
Okwi, P.O., Godfrey, N., Patti, K., Mike, A., An, N., Abisalom, O., Norbert, H., Todd, B.,
Patrick, K., John, O., 2007. Proc. Natl. Acad. Sci. U. S. A. 104, 16769–16774. https://
doi.org/10.1073/pnas.0611107104.
Porterfield, S.L., McBride, T.D., 2007. The effect of poverty and caregiver education on
perceived need and access to health services among children with special health care
needs. Am. J. Publ. Health 97, 323. https://doi.org/10.2105/AJPH.2004.055921.
Wang, Y., Wang, M., Wu, D., Zhao, L., Ding, J., 2017. Spatial patterns and determinants of
rural poverty: a case of Guizhou province, China. Sci. Geogr. Sin. 37, 217–227. (in
Chinese). http://www.wanfangdata.com.cn/details/detail.do?_type=perio&id=
dlkx201702007.
Wu, Y., Feng, W., Zhou, Y., 2019. Practice of barren hilly land consolidation and its
impact: a typical case study from Fuping county, Hebei province of China. J. Geogr.
Sci. 29, 762–778. https://doi.org/10.1007/s11442-019-1626-x.
Xian, Z., Wang, P., Wu, W., 2016. Rural poverty lines and poverty monitoring in China.
Statistical Research 33, 3–12. (in Chinese). http://www.cnki.com.cn/Article/
CJFDTotal-TJYJ201609001.htm.
Xu, W., Yin, Y., Zhou, S., 2007. Social and economic impacts of carbon sequestration and
land use change on peasant households in rural China: a case study of Liping,
Guizhou Province. J. Environ. Manag. 85, 736–745. https://doi.org/10.1016/j.
jenvman.2006.09.013.
Yang, R., Xu, Q., Long, H., 2016. Spatial distribution characteristics and optimized re-
construction analysis of China's rural settlements during the process of rapid urba-
nization. J. Rural Stud. 47, 413–424. https://doi.org/10.1016/j.jrurstud.2016.05.
013.
Yang, W., Li, T., Cao, X., 2015. The spatial pattern of Community Travel Low Carbon
Index (CTLCI) and spatial heterogeneity of the relationship between CTLCI and in-
fluencing factors in Guangzhou. Geogr. Res. 34, 1471–1480. (in Chinese). http://
www.cnki.com.cn/Article/CJFDTotal-DLYJ201508007.htm.
Zhang, X., 2016. History, achievements, problems and suggested countermeasures of
karst desertification control in Guizhou. Carsol. Sin./Zhong Guo Yan Rong 35,
497–502. (in Chinese). http://en.cnki.com.cn/Article_en/CJFDTotal-
ZGYR201605005.htm.
Zhang, X., Chen, S., Liao, C., Song, J., 2016. Spatial spillover effects of regional economic
growth in Beijing-Tianjin-Hebei region. Geogr. Res. 35, 1753–1766. (in Chinese).
http://en.cnki.com.cn/Article_en/CJFDTOTAL-DLYJ201609014.htm.
Zhou, L., Xiong, L., 2018. Natural topographic controls on the spatial distribution of
poverty-stricken counties in China. Appl. Geogr. 90, 282–292. https://doi.org/10.
1016/j.apgeog.2017.10.006.
Zhou, L., Zhen, F., Wang, Y., Xiong, L., 2019. Modeling the spatial formation mechanism
of poverty-stricken counties in China by using geographical detector. Sustain. Times
11, 4752. https://doi.org/10.3390/su11174752.
Zhou, Y., Liu, Y., 2019. The geography of poverty: review and research prospects. J. Rural
Stud. https://doi.org/10.1016/j.jrurstud.2019.01.008.
J. Xu, et al. Journal of Rural Studies xxx (xxxx) xxx–xxx
10
