EN中文

Journal of Resources and Ecology >

2021 , Vol. 12 >Issue 2: 192 - 202

DOI: https://doi.org/10.5814/j.issn.1674-764x.2021.02.006

Land Use Change and Land Multifunction Tradeoffs

Spatial Behavior Characteristics of Land Use based on Fractal Theory: Taking Poyang Lake Area as an Example

  • HE Yafen , *
Expand
  • Institute of Ecological Civilization, Jiangxi University of Finance and Economics, Nanchang 330013, China
*: HE Yafen, E-mail:

Received date: 2020-09-30

  Accepted date: 2020-12-28

  Online published: 2021-05-30

Supported by

The National Natural Science Foundation of China(41961035)

The Natural Science Foundation in Jiangxi Province(20202BAB213014)

The China Postdoctoral Science Foundation(2020M682106)

Fold

Abstract

Landscape morphology can reflect the spatial behavior of land use. Using the Poyang Lake area as an example, the landscape pattern characteristics in 1995, 2000, 2015, and 2018 are determined by calculating the fractal dimension, fractal stability, patch density, patch shape fragmentation, and landscape isolation, and fractal theory is used to analyze the spatial behavior of land use. The results show that building land was the land use type which consistently had the highest fractal dimension, but the fractal dimension of building land shows a downward trend, indicating that the spatial form of building land gradually developed in an orderly direction under the action of land use spatial behavior. Paddy, dryland, and forested land were the land use types which always had the lowest fractal dimension, and they are in unstable states. The calculation results of patch density, patch shape fragmentation index, and landscape isolation index supported the conclusions of the fractal analysis. One recommendation for realizing the rational layout of the land is to reduce the fractal dimension of building land through scientific and reasonable planning and to guide the orderly development of building land. For natural landscapes such as forested land, shrub forest land, high-coverage grassland, and water area, their fractal dimensions should be increased to reduce human interference and maintain their stability. Finally, the results of this study suggest that the fractal dimension should be introduced into the National Spatial Planning, and used as an index for evaluating the rationality of the regional land use pattern.

Key words： land use; fractal theory; landscape pattern; Poyang Lake area

Cite this article

HE Yafen . Spatial Behavior Characteristics of Land Use based on Fractal Theory: Taking Poyang Lake Area as an Example[J]. Journal of Resources and Ecology, 2021 , 12(2) : 192 -202 . DOI: 10.5814/j.issn.1674-764x.2021.02.006

1 Introduction

In conjunction with the rapid development of urbanization, regional land use has changed greatly, especially building land and farmland. Many studies have analyzed the area changes and location changes of land use at different levels (Lv et al., 2017; Liu et al., 2018; Tan and Li, 2019; Kong et al., 2020; Wu et al., 2020), but they have largely ignored the landscape changes in the ecological sense.
Landscape ecology is a discipline which studies spatial heterogeneity (Xu et al., 2001), and the spatial pattern is an important component of landscape ecology research. Therefore, fractal theory, which is known for analyzing spatial structure, is being increasingly applied to landscape ecological studies, such as those focusing on the complexity and stability of landscape mosaic structures (Xie and Li, 2008; Shi et al., 2020). Fractal theory was developed in the 1970s and it spans the natural sciences, social sciences, and thinking sciences. It provides a powerful quantitative method for the study of complex regional landforms and land use patterns (Mandelbrot, 1975; Mandelbrot, 1983; Barnsley, 1988; van Hees, 1994).
Since the 1990s, fractal theory has been widely used in geography in China. Great progress has been made in urban geography, cartography, and remote sensing, among which the most widely used and most mature tool is the landscape patch area-perimeter model (Xiao et al., 1997; Xu et al., 2001). For example, Tian and Ren (2012) used the fractal model to conduct an empirical study on the spatial change of land use in Baota District, which allowed them to elaborate on the temporal and spatial variations of different land use types based on the land use data of Baota District of Yan’an City in 1997 and 2002 with the support of GIS (Geographic Information System). Pan et al. (2015) analyzed the growth and morphological evolution of major cities in China using the compactness index, the shape index, and the fractal dimension. Liang et al. (2017) analyzed the spatial evolution of rural settlements in Xiamen from 1986 to 2016 with the help of a fractal model, and proposed strategies for optimizing the spatial structure of rural settlements based on the characteristics of their spatial differences. Yu et al. (2019) conducted a quantitative study on the boundary complexity of Xiamen based on fractal theory. Xu et al. (2019) extracted the land use change characteristics of Neijiang City using a fractal model. Ren (2019) used a fractal model to analyze the changes in the urban form of Kunming. Such recent studies have mostly taken a specific landscape as the research object to study the landscape self-similarities, such as rural settlements, cities (Bosch et al., 2020; Song et al., 2020), or water (Mirzaei et al., 2020), but a few studies have focused on the whole regional land use situation and analyzed the landscape changes to reflect the spatial behavior of land use. As a result, it is difficult to use the results of fractal theory analysis to guide the development of regional land patterns.
As the largest freshwater reservoir in China, Poyang Lake is an important ecological function area. What changes have taken place in the landscape structure of Poyang Lake area in the past 20 years? What is the direction of regional planning based on sustainable landscape ecology? This research uses fractal theory to analyze the spatial behavior of land use. With the support of RS and GIS, the land use in the Poyang Lake area is taken as the research object. The theory and research methods of landscape ecology, especially fractal theory, are used to analyze the landscape pattern changes in order to explore the laws governing the evolution of the landscape pattern in Poyang Lake area.

2 Data sources and research methods

2.1 Study area

The Poyang Lake area (28°22′-29°45′N, 115°47′-116°45′E) is located on the southern bank of the middle and lower reaches of the Yangtze River and the northern part of Jiangxi Province (Fig. 1). It is an important ecological base for food, oil, cotton, and fish. The total area is 19731 km2, the GDP in 2019 was 355.414 billion yuan, and the population is about 7.7 million.
Poyang Lake is the largest freshwater lake and one of the most important ecological function protection areas in China. It plays important roles in water conservation, soil conservation, flood regulation, and biodiversity conserva tion to maintain regional and national ecological security.
Fig. 1 Location of Poyang Lake area

2.2 Data sources and methods

The data used in this paper are on the classified land use status in 1995, 2005, 2015, and 2018, which are sourced from the school of Geography and Science Planning, Sun Yat-sen University. The main land use types include paddy, dryland, forested land, shrub forest land, sparse forest land, other forest land, high-coverage grassland, low-medium- coverage grassland, water area, building land, and unused land.
The fractal refers to a shape whose component parts are similar to the whole in some form, and two important characteristics of the fractal are self-similarity and scale invariance (Burrough, 1986; Edgar, 1990; Falconer, 1990; Falconer, 1997). The self-similarity of a system refers to “the characteristics of a result or process are similar on different time or space scales”, and the scaling invariance means that “the self-similarity still exists as the scale changes”. The fractal dimension is one of the quantitative indicators that characterize self-similar structures or systems. It describes the complexity of a shape by quantitatively describing the size of the core area and the tortuosity of the boundary line.
The landscape shapes of land use are derived from complex geographic phenomena and show the typical fractal performance. The fractal dimension is the organic combination of the area, perimeter and number of land use patches, which can express the hidden information that the traditional statistical methods of area and perimeter cannot. The value of the fractal dimension is generally between 1 and 2. For a land patch fractal, the larger the fractal dimension, the longer the perimeter of the land patch with the same area, and the more complex the land patch. An increase of the fractal dimension indicates that the spatial mosaic structure has become more complicated.
At the same time, land use type is a product of the dual effects of natural and human activities. It has the characteristics of irregularity, relative instability, and complexity. Therefore, it can be explored using the stability index in the fractal method. The stability index is directly related to the fractal dimension, and can be used to characterize the spatial behavior of land use. The original meaning of the stability index is to show the ability of a land patch with a certain fractal dimension to resist external interference and maintain its form in the natural state. According to the relationship between the stability index and the fractal dimension, the landscape is most unstable when the fractal dimension is 1.5; when the fractal dimension is greater than 1.5, the fractal dimension of the landscape and its stability index change in the same direction; but when the fractal dimension is less than 1.5, the fractal dimension of the landscape and its stability index change in opposite directions.
Landscape pattern indexes such as patch density, patch shape fragmentation, and landscape isolation can also reflect the spatial behavior of land use. Patch density describes the fragmentation of the landscape. Patch shape fragmentation represents the degree of fragmentation of the landscape and reflects the complexity of the spatial structure of the landscape and the degree of human interference with the landscape. Landscape isolation refers to the degree of isolation among different patches in a certain landscape type. To some extent, the degree of isolation reflects the influence of human activity intensity on landscape structure. Therefore, the values of patch density, patch shape fragmentation, and landscape isolation can more comprehensively reflect the evolution of regional land use spatial behavior characteristics.
Fractal theory is used to study land use spatial behavior, mainly focusing on the changes of landforms caused by human land use spatial behavior and providing an explanation. The analysis of other indicators that represent the results of land spatial behavior also follows this idea, that is, it pays attention to the change directions and amounts of index values, and explains its behavior.
The relevant indicators, the calculation method, and the significance of the parameters in the model are as follows:
(1) Fractal Dimension
The perimeter-area model used to calculate the fractal dimension is:
$\ln \frac{P}{4}=k\ln A+c,\begin{matrix} {} \\ \end{matrix}FD=2k$
where P is the patch perimeter; A is the patch area; k is the slope of the regression equation; c is the constant; and FD represents the “average” fractal dimension of a landscape containing multiple patches, which is also the statistical fractal dimension of the landscape. The theoretical range of FD is [1.0, 2.0], where FD=1.0 represents the simplest square patch, and FD=2.0 represents the most complex perimeter patch over the same area.
(2) Landscape Stability Index
For a certain landscape element, the higher the FD value, the more complex the mosaic structure of that element; when FD=1.5, it means that the landscape element is in a random state similar to Brownian motion, that is, an unstable state; and the closer the FD value is to 1.5, the more unstable the factor is (Xu et al., 2001). Therefore, the stability index (SK) of each landscape element can be defined as:
$SK=\left| 1.5-FD \right|$
(3) Patch Shape Fragmentation Index
$FS=1-\frac{1}{ASI}$
$ASI=\sum\limits_{i=1}^{n}{\frac{A(i)SI(i)}{A}}$
$SI(i)=\frac{P(i)}{4\sqrt{A(i)}}$
$A=\sum\limits_{i=1}^{n}{A(i)}$
where, FS is the patch shape fragmentation index of a certain landscape type, and ASI is the average patch shape index weighted by area; SI(i) is the shape index of landscape patch i, and P(i) is the perimeter of landscape Patch i; A(i) is the area of landscape patch i and A is the total area of the landscape type; and n is the number of patches of the landscape type.
(4) Patch Density
The patch density of the landscape represents the number of patches per unit area including all patches of heterogeneous elements in the landscape, and the calculation formula is:
$P{{D}_{i}}=\frac{{{N}_{i}}}{A(i)}$
where PDi is the patch density of landscape type i; Ni is the number of patches of landscape type i; A(i) is the area of the landscape type i.
(5) Landscape Isolation Index
${{F}_{i}}=\frac{{{D}_{i}}}{{{S}_{i}}}$
${{D}_{i}}=\frac{1}{2}\sqrt{\frac{A}{{{N}_{i}}}}$
${{S}_{i}}=\frac{A(i)}{A}$
where, Fi is the separation degree of a certain landscape type; Di is the distance index of landscape type i; Si is the area index of landscape type i; A(i) is the area of landscape type i and A is the total area of the study area; and Ni is the number of patches of landscape type i.

3 Results and analysis

3.1 Changes of land use

The results of land use change in the Poyang Lake area from 1995 to 2018 obtained by using GIS are shown in Fig. 2. Table 1 shows that the land use in the Poyang Lake area in the past 25 years (from 1995 to 2018) has changed greatly, especially the expansion of building land. The building land area was 498.39 km2 in 1995, but it doubled to 1000.76 km2 by 2018. In contrast, the areas of paddy, dryland, shrub forest land, sparse forest land, high-coverage grassland, and low-medium-coverage grassland all decreased during this period. For example, the area of paddy decreased by 228.41 km2 and dryland decreased by 118.22 km2. In addition, the water area increased slightly. As Fig. 2 shows, the areas of expansion of building land are mainly distributed in the southwestern and northern parts of the region, and the building land along the Poyang Lake was gradually cleared.
Table 1 Land use type areas and composition ratios in 1995, 2005, 2015 and 2018
Land type 1995 2005 2015 2018
Area (km2) Percentage (%) Area (km2) Percentage (%) Area (km2) Percentage (%) Area (km2) Percentage (%)
Paddy 6972.09 35.34 6827.96 34.61 6788.04 34.40 6743.68 34.18
Dryland 1991.61 10.09 1955.08 9.91 1912.14 9.69 1873.38 9.49
Forested land 2707.73 13.72 2749.59 13.94 2901.73 14.71 2884.44 14.62
Shrub forest land 895.04 4.54 870.69 4.41 775.95 3.93 774.94 3.93
Sparse forest land 1326.84 6.72 1286.09 6.52 1143.53 5.80 1127.08 5.71
Other forest land 27.18 0.14 34.14 0.17 36.52 0.19 36.24 0.18
High-coverage grassland 354.35 1.80 331.52 1.68 299.40 1.52 298.79 1.51
Low-medium-coverage grassland 345.33 1.75 350.22 1.77 321.80 1.63 320.98 1.63
Water 4609.34 23.36 4708.54 23.86 4676.89 23.70 4670.10 23.67
Building land 498.39 2.53 614.71 3.12 873.40 4.43 1000.76 5.07
Unused land 3.10 0.02 2.48 0.01 1.61 0.01 0.60 0
Total 19731.00 100.00 19731.00 100.00 19731.00 100.00 19731.00 100.00

3.2 Fractal dimension and stability analysis

(1) Fractal dimension of land use
The results of the fractal dimensional analysis of the different land use types in 1995, 2005, 2015, and 2018 are shown in Table 2. Each land type may contain different land use actors, and each land use actor may have one or more land patches. This study does not discuss the fractal situation of a single land patch, nor all the land patches of a single land use actor, but mainly focuses on whether the distribution of a given land use type has fractal characteristics. The fractal dimension of a given type of land use is the statistical “average” fractal dimension of multiple land patches of the same type.
Fig. 2 The trends of land use change in the Poyang Lake area in 1995, 2005, 2015, and 2018.
The specific method used here substitutes the area and perimeter data of each patch into the logarithmic formula, generates a scatter plot for each data pair, and then performs a linear simulation on all the scattered points. The slope of the linear equation is used to find the fractal dimension of a given type of land. Figure 3 shows the scatter diagrams of the area and perimeter of paddy after taking the logarithm in the Poyang Lake area in 1995, 2005, 2015, and 2018.
The same method was applied to the other land use types in the Poyang Lake area. Table 2 list the regression equations, coefficients of determination, and sample numbers of the patches of various land use types in the Poyang Lake area in 1995, 2005, 2015, and 2018. The fractal dimensions of the various land use types are shown in Table 3.
Fig. 3 Scatter diagrams of the area and perimeter of paddy after taking logarithm in Poyang Lake area in 1995, 2005, 2015, and 2018.
Based on the results of the fractal dimension calculations (Table 3), the patch distribution of various land use types in the Poyang Lake area has good self-similarity, so it is feasible to use fractal theory to analyze the landscapes in general.
Comparing the fractal dimensions of various land use types in 1995, 2005, 2015, and 2018, the building land, forested land, and water area were reduced from 1995 to 2005, while the fractal dimensions of other land use types increased. Among them, building land endured the largest decrease (0.0193) of its fractal dimension, and other forest land had the largest increase (0.0494). From 2005 to 2015, the fractal dimensions of low-medium-coverage grassland and building land decreased, while those of the other land use types increased. Among them, low-medium- coverage grassland endured the largest decrease of its fractal dimension (0.0086), and other forested land had the largest increase (0.0701). From 1995 to 2015, the fractal dimensions of building land and forested land decreased, while those of the other land use types increased. From 2015 to 2018, building land had the largest increase (0.011) of its fractal dimension. In the past 25 years, other forest land had the largest increase in fractal dimension (0.0119), while building land endured the largest decrease (0.011).
Table 2 Regression equations and related statistics of the fractal calculations of various land use types in Poyang Lake area in 1995, 2005, 2015, and 2018.
Land type Year Regression equation Number of samples
Paddy 1995 lnA = 0.0708+1.5173lnP (R2=0.9746) 7860
2005 lnA = 0.0436+1.5224lnP (R2=0.9748) 8034
2015 lnA = 0.0124+1.528lnP (R2=0.9746) 8115
2018 lnA = 0.0201+1.527lnP (R2=0.9746) 7987
Dryland 1995 lnA = ‒0.2253+1.5802lnP (R2=0.9729) 5975
2005 lnA = ‒0.2253+1.5805lnP (R2=0.9728) 6061
2015 lnA = ‒0.2909+1.5905lnP (R2=0.9728) 6093
2018 lnA = ‒0.2872+1.5901lnP (R2=0.9726) 5960
Forested land 1995 lnA = ‒0.2403+1.5828lnP (R2=0.9741) 4415
2005 lnA = ‒0.2168+1.5799lnP (R2=0.9742) 4461
2015 lnA = ‒0.2295+1.5822lnP (R2=0.974) 4501
2018 lnA = ‒0.2244+1.5816lnP (R2=0.974) 4432
Shrub forest land 1995 lnA = ‒0.2971+1.5902lnP (R2=0.9766) 2221
2005 lnA = ‒0.305+1.5915lnP (R2=0.977) 2244
2015 lnA = ‒0.4031+1.6055lnP (R2=0.9763) 2236
2018 lnA = ‒0.4098+1.6065lnP (R2=0.9765) 2237
Sparse forest land 1995 lnA = ‒0.7041+1.6497lnP (R2=0.9733) 5692
2005 lnA = ‒0.7117+1.6514lnP (R2=0.9729) 5629
2015 lnA = ‒0.7388+1.6564lnP (R2=0.9687) 5472
2018 lnA =‒0.7429+1.657lnP (R2=0.9725) 5394
Other forest land 1995 lnA = ‒0.2654+1.6039lnP (R2=0.9599) 136
2005 lnA = ‒0.6172+1.6533lnP (R2=0.9632) 183
2015 lnA = ‒1.1316+1.7234lnP (R2=0.9482) 189
2018 lnA = ‒1.1228+1.7226lnP (R2=0.9657) 187
High-coverage grassland 1995 lnA = ‒0.4133+1.612lnP (R2=0.9754) 765
2005 lnA = 0.4389+1.6152lnP (R2=0.9754) 836
2015 lnA = ‒0.4624+1.6209lnP (R2=0.9639) 843
2018 lnA = ‒0.4662+1.6214lnP (R2=0.975) 842
Low-medium-coverage grassland 1995 lnA = ‒0.3172+1.6076lnP (R2=0.972) 538
2005 lnA = ‒0.3962+1.6166lnP (R2=0.9735) 559
2015 lnA = ‒0.3234+1.608lnP (R2=0.9736) 575
2018 lnA = ‒0.3298+1.609lnP (R2=0.9736) 576
Water area 1995 lnA = ‒0.5243+1.6132lnP (R2=0.9736) 3900
2005 lnA = ‒0.4799+1.6066lnP (R2=0.9746) 3884
2015 lnA = ‒0.5465+1.6177lnP (R2=0.9749) 3920
2018 lnA = ‒0.5443+1.6173lnP (R2=0.9749) 3898
Building land 1995 lnA = ‒1.3431+1.7532lnP (R2=0.9506) 5303
2005 lnA = ‒1.2109+1.7339lnP (R2=0.9536) 5300
2015 lnA = ‒1.2066+1.7333lnP (R2=0.9592) 5325
2018 lnA = ‒1.2675+1.7422lnP (R2=0.961) 5288
Table 3 Comparison of fractal dimensions and stabilities of various land use types in the Poyang Lake area in 1995, 2005, 2015 and 2018.
Land type Fractal dimension Stability index Stability ranking
1995 2005 2015 2018 1995 2005 2015 2018 1995 2005 2015 2018
Building land 1.753 1.734 1.733 1.742 0.253 0.234 0.233 0.242 1 1 1 1
Sparse forest land 1.650 1.651 1.656 1.657 0.150 0.151 0.156 0.157 2 3 3 3
High-coverage grassland 1.612 1.615 1.620 1.621 0.112 0.115 0.120 0.121 4 5 4 4
Water area 1.613 1.607 1.618 1.617 0.113 0.107 0.118 0.117 3 6 5 5
Low-medium-coverage grassland 1.608 1.617 1.608 1.609 0.108 0.117 0.108 0.109 5 4 6 6
Other forest land 1.604 1.653 1.724 1.723 0.104 0.153 0.224 0.223 6 2 2 2
Shrub forest land 1.590 1.592 1.606 1.607 0.090 0.092 0.106 0.107 7 7 7 7
Forested land 1.583 1.580 1.582 1.581 0.083 0.080 0.082 0.081 8 9 9 9
Dryland 1.580 1.581 1.591 1.590 0.080 0.081 0.091 0.090 9 8 8 8
Paddy 1.517 1.522 1.528 1.527 0.017 0.022 0.028 0.027 10 10 10 10
In 1995, 2000, 2015, and 2018, there were no major natural disasters in the Poyang Lake area, so most of the changes in land use form and landscape patterns were caused by human land use spatial behavior. Based on fractal theory, the larger the fractal dimension, the more complex the land patch. A decrease of the fractal dimension means that the shape of the land patch has become more regular. For production, the more regular the shape of a land patch, the more conducive it is to production efficiency, due to the reductions of transportation costs and the empty rate of field machinery. Generally speaking, orderly land use behavior will make the land patch more regular and reduce the fractal dimension.
For the land use types with a reduced fractal dimension, the combination or division of the space causes these land use types to develop in an orderly direction. The decrease of the fractal dimension not only shows the improvement of land patch morphology regularity but also the intensity of human behavior in improving the landform, such as afforestation or returning farmland to forest. For the land types in which the fractal dimension increased, the spatial behavior of land use made the land patch shape develop in the direction of fragmentation, and the disorder of land use increased, such as the disorderly expansion of rural building land. For all land, the change of the landform and landscape pattern is directly related to the stimulation of land use behavior by the economic development policy of the Poyang Lake area, and it is the reflection of social-economic policy in the space of land utilization.
(2) Stability analysis
The stability index directly relates to the fractal dimension and represents the result of land use spatial behavior. By exploring the relationship between it and the fractal dimension, the change in stability and development direction of the land patch caused by the change of the fractal dimension can be clarified, and it can be used to evaluate the spatial behavior of land use. In this regard, it is helpful to put forward countermeasures and suggestions that can control the spatial behavior of land use from the perspective of planning to achieve better stability of the landscape.
In Table 3, although the fractal dimension of building land decreased on the temporal scale, the stability index of building land was always ranked first, making it the most stable land use type. At the other end of the spectrum, the stability index of paddy was always at the bottom. Compared with other types of land, paddy generally has a more regular shape to facilitate water drainage and irrigation. Therefore, its fractal dimension is relatively low. Also, the fractal stability of other forest land shows a great change from 1995 to 2018, which indicates a great improvement from the sixth rank to the second rank. The fractal stability of the other land use types changed very little.

3.3 Change characteristics of the landscape index

(1) Analysis of patch density
Patch density represents the number of patches per unit area, and it is quantified in this paper as the number of patches per hectare. In general, the greater the patch density of a certain type of land, the more patches per unit area of that type of land. The smaller the average size of each patch, the more severe the patch segmentation, and the smaller the patch density. On the contrary, patch density to some extent also reflects the fragmentation of land patches. The patch densities of the various land use types in the Poyang Lake area in 1995, 2005, 2015, and 2018 are shown in Table 4.
The patch density of forested land, water area, and building land decreased from 1995 to 2005, while the patch density of other land use types increased. From 2005 to 2015, the land use types with a reduced patch density include forested land, other forest land, and building land, and the patch density of other land use types increased. From 2015 to 2018, the patch density of the shrub forest land, high-coverage grassland, and low-medium-coverage grassland decreased, while that of the other land use types increased. According to the change over 20 years, paddy, dryland, shrub forest land, sparse forest land, other forest land, high-coverage grassland, low-medium-coverage grassland, and water area all increased in patch density, while forested land and building land decreased in patch density. A decrease in patch density indicates that land patches are merged more than divided, and the average area of the land patches increases; but with the increase of patch density, the segmentation of patches is more obvious, the average area of the land patches is reduced, and the degree of fragmentation of the patches increases. Comparing the fractal dimension and patch density data of various types of land in the Poyang Lake area, indicates that the changing direction of the land patch density and the changing direction of the fractal dimension are the same, which further verifies that the spatial segmentation and merging of land use behavior subjects is the fundamental cause of landform change.
(2) Analysis of patch shape fragmentation
Similar to the fractal dimension, the patch shape fragmentation index is an important parameter for describing the landscape characteristics in landscape ecology, which can further explain the generality of the results of land use spatial behavior in the Poyang Lake area. The adoption of other landscape ecology indicators also follows this principle, which is for supplemental and additional verification of the results of fractal theory. Table 4 shows the patch shape fragmentation index for the land use types in the four periods of 1995, 2005, 2015, and 2018 in the Poyang Lake area.
Comparing the fragmentation index values of the various land use types in the Poyang Lake area in 1995, 2005, 2015, and 2018 in Table 4, the patch fragmentation of the various land types increased from 1995 to 2005, except for forested land, water area, and building land. From 2005 to 2015, the patch fragmentation of the different land types increased, except for forested land, other forest land, and building land. From 2015 to 2018, the patch fragmentation of paddy, dryland, forested land, other forest land, water area, and building land decreased. In the 20 years from 1995 to 2018, the patch fragmentation of building land decreased dramatically, indicating that building land had gradually coalesced from a very disjointed state, mainly due to the fact that the number of scattered rural settlements had decreased, while the range of urban building land had expanded. In addition, the water area and forested land had a slight decrease in patch fragmentation. But the patch fragmentation of other land use types, such as paddy, dryland, shrub forest land, and high-coverage grassland, all increased, indicating that these land use types were increasingly subjected to human interference.
(3) Analysis of landscape isolation
Landscape isolation reflects the influence of human activities on landscape structure to a certain extent. The data in Table 4 show that the isolation index of dryland, shrub forest land, sparse forest land, other forest land, high-coverage grassland, low-medium-coverage grassland, and building land is very large. The reason for such isolation is that Poyang Lake is a huge water body and the most important land use type in the region. Since the other land types are all distributed around the huge water body of Poyang Lake, this makes the separation index of these land types very large. From 1995 to 2005, the isolation of paddy, dryland, shrub forest land, sparse forest land, and high-coverage grassland increased, while the isolation of the other land types decreased, which shows that the distribution of these land types became more dispersed during this period. From 2005 to 2015, the landscape isolation of forest land, shrub forest land, sparse forest land, high-coverage grassland, and low-medium-coverage grassland increased, while for the other land types the isolation decreased. From 2015 to 2018, the landscape isolation of other forest land, water area, and building land decreased, while it increased for the other land types. In the 20 years from 1995 to 2018, the most obvious decrease in the landscape isolation was that of building land, possibly due to the rapid expansion of urban building land.
Table 4 Comparisons of landscape indices of various land use types in the Poyang Lake area in 1995, 2005, 2015, and 2018.
Land type Patch density Patch shape fragmentation index Landscape isolation index
1995 2005 2015 2018 1995 2005 2015 2018 1995 2005 2015 2018
Paddy 0.115 0.122 0.134 0.133 0.327 0.354 0.389 0.388 102.9 112.0 102.3 103.8
Dryland 0.175 0.180 0.185 0.181 1.738 1.811 1.908 1.904 40585.2 41653.0 40985.8 42553.7
Forested land 0.138 0.139 0.138 0.136 1.006 0.997 0.941 0.933 1761.1 1737.8 2298.6 2299.1
Shrub forest land 0.068 0.069 0.070 0.070 1.490 1.562 1.768 1.773 99800.8 103420.2 121972.1 122214.4
Sparse forest land 0.196 0.198 0.193 0.191 2.915 3.038 3.335 3.343 133032.9 145093.5 209645.8 212327.8
Other forest land 0.006 0.008 0.008 0.008 4.039 4.574 4.302 4.280 18062511.6 16460734.6 15977976.6 16021729.2
High-coverage grassland 0.026 0.028 0.029 0.029 1.444 1.662 1.897 1.917 127222.9 145349.2 209863.9 209928.1
Low-medium-coverage grassland 0.021 0.021 0.022 0.023 1.184 1.203 1.374 1.403 348843.5 301706.5 415720.3 416993.9
Water area 0.106 0.104 0.107 0.106 0.455 0.437 0.452 0.450 25.5 24.3 24.9 25.1
Building land 0.242 0.240 0.240 0.239 9.587 7.712 5.431 4.719 343695.5 105461.8 28319.9 14198.1
Overall landscape pattern 1.094 1.110 1.127 1.116 1.094 1.110 1.127 1.116 20.2 19.7 19.8 19.9
According to Table 4, the patch density of each kind of land in the Poyang Lake area changed from 1995 to 2018, and the patch density index of the overall landscape showed an increasing trend. Except for building land, water area, and sparse forest land, the patch density index values of other land types increased. During the study period, the fragmentation index of the regional landscape patches increased from 1.094 to 1.116, indicating that the fragmentation of regional land use landscape patterns decreased. Among all of the land types, the most changeable were building land, forested land, and water area. From the perspective of the landscape isolation index, the overall degree of isolation of the Poyang Lake landscape decreased from 20.2 in 1995 to 19.7 in 2005, and then it increased to 19.9 in 2018. As far as individual types are concerned, the land use types for which the degree of isolation has been greatly reduced include other forest land, water area, and building land, among which the decline of building land is the most obvious. The reason is that the expansion of urban building land tends to expand to the peripheral base on the original building land. At the same time, the relocation of rural residential areas reduces the previously scattered rural residential areas over time and increases the degree of agglomeration of the building land. The main reason for the increases in the isolation degree of other land types is the enhancement of human disturbance intensity.

4 Discussion

Landscape morphology can reflect the spatial behavior of land use. For the land use types with a fractal dimension which was reduced, the combination or division of the land use behavior subjects made them develop in an orderly direction. The decrease of the fractal dimension indicates an improvement of the land patch morphology. For example, Pan et al. (2015) found that under the control of land use planning and urban planning, the outline of the expansion of built-up areas in Chinese major cities from 1990 to 2010 tended to be regular and tidy, the land use was more compact and intensive, and these features were manifested in the reduction of the fractal dimension of the city. Therefore, existing research has concluded that the land fractal dimension and stability index can better serve land use planning and provide practical guidance for land development and land consolidation (Tian and Ren, 2012).
That is to say, the fractal dimension should be reduced for land types with higher levels of manual participation from the perspective of planning. Thus, based on the current situation of land fractal characteristics in the Poyang Lake area, the fractal dimension of regional building land is still very high, and reasonable and scientific planning should be adopted to make it more orderly and regular.
Previous land fractal studies have not given clear guidance on the pattern optimization of ecological lands, such as forested land and water areas. As the intensity of human disturbance increases, many ecological land types are disturbed, resulting in the reduction of their biodiversity. An increase of the fractal dimension indicates an increase of the perimeter of a land patch with the same area, and that the land patch has become more complicated. Therefore, the results of this study indicate that increasing the fractal dimension of natural ecological land and its stability is more conducive to the protection of natural ecological land.
Spatial structure and spatial form together constitute the spatial characteristics of the land use landscape, and the scientific analysis of both of these provides the basis for land use pattern optimization. The application of fractal theory to the analysis of both land use landscape spatial pattern changes and its mosaic structure is an important tool for explaining the changes in the characteristics of land use spatial behavior, as well as for landscape evaluation, management, and regional sustainable development. At present, the National Spatial Planning emphasizes not only the layout of urban development land but also the rational planning of ecological land. Therefore, one recommendation is to introduce the fractal dimension into the National Spatial Planning, and to use it as an indicator for evaluating the rationality of the regional land use pattern and for introducing the landscape pattern thinking into the regional land space planning process.
The introduction of fractal theory into the study of the land use spatial behavior is only the first step, and how the changes of the fractal dimension further affect the ecological process of the regional landscape remains unknown and needs further exploration.

5 Conclusions

Based on the description of the land use status of the Poyang Lake area in 1995, 2000, 2015, and 2018, this study first used the fractal model to calculate the fractal dimension and stability index of various land-use types and analyzed their changing trends. Then, the three landscape pattern indexes of patch density, patch shape fragmentation index, and landscape isolation index were used to further explain the changes in the regional land use landscape, to verify the results of fractal theory, and to further illustrate the generality of land use spatial behavior in the Poyang Lake area.
The results show that during the past 20 years from 1995 to 2018, building land was always the land use type with the highest fractal dimension, while paddy was always the land use type with the lowest fractal dimension. However, from the perspective of the fractal dimension change, the changes of building land and forested land showed downward trends, which were not obvious, while the fractal dimensions of the other land types increased. Because the fractal dimensions of all the different kinds of land were greater than 1.5, the changes of their fractal dimensions also indicate the changes of stability index values. The analysis results of the patch density, patch shape fragmentation index, and landscape isolation index can verify the results of fractal theory. The decrease of the patch fragmentation index of forest land was the main reason for the decline of the fractal dimension of this type of land. Meanwhile, the decreases of the patch density, patch shape fragmentation index, and landscape isolation index caused the decrease of the fractal dimension of building land, and the increase of the patch density of water area was the main reason for the increase of its fractal dimension.
Finally, the derived results have been confirmed by comparing them to those reported in articles with similar themes, and more insights that fractal theory may be able to provide in landscape research are discussed.

References

Publishing order | Descend order by publishing year | Descend order by cited within
1
Barnsley M F. 1998. Fractals everywhere. London, UK: Academic Press.

2
Bosch M, Jaligot R, Chenal J . 2020. Spatiotemporal patterns of urbanization in three Swiss urban agglomerations: Insights from landscape metrics, growth modes and fractal analysis. Landscape Ecology, 35(4): 879‒891.

3
Burrough P A . 1986. Principles of geographical systems for land resources assessment. Oxford, UK: Clarendon.

4
Edgar G A . 1990. Measure, topology and fractal geometry. New York, USA: Springer-Verlag.

5
Falconer K J . 1990. Fractal geometry: Mathematical foundations and applications. Chichester, UK: John Wiley & Sons.

6
Falconer K J . 1997. Techniques in fractal geometry. Chichester, UK: John Wiley & Sons.

7
Kong X L, Li Y L, Han M , et al. 2020. Analysis of land use/cover change and landscape pattern in the Yellow River Delta during 1986‒2016. Journal of Southwest Forestry University, 40(4):122-131. (in Chinese)

8
Liang F C, Liu S Y, Liu L M . 2017. Spatial characteristics and evolution of rural settlement landscape based on fractal theory: A case study of Xiamen, China. Chinese Journal of Applied Ecology, 28(8): 2640‒2648. (in Chinese)

9
Liu J Y, Ning J, Kuang W H , et al. 2018. Spatio-temporal patterns and characteristics of land-use change in China during 2010‒2015. Acta Geographica Sinica, 73(5): 789‒802. (in Chinese)

10
Lv L G, Li Y L, Sun Y . 2017. The spatio-temporal pattern of regional land use change and eco-environmental responses in Jiangsu, China. Journal of Resources and Ecology, 8(3): 268‒276.

11
Mandelbrot B B . 1975. Stochastic models for the Earth’s relief, the shape and the fractal dimension of the coastlines, and the number-area rule for islands. Proceedings of the National Academy of Sciences of the USA, 72(10): 3825‒3828.

PMID

12
Mandelbrot B B . 1983. The Fractal geometry of mature. New York, USA: W. H. Freeman and Company.

13
Mirzaei M, Jafari A, Gholamalifard M , et al. 2020. Mitigating environmental risks: Modeling the interaction of water quality parameters and land use cover. Land Use Policy, 95(6):103766. DOI: 10.1016/j.landusepol.2018.12.014.

14
Pan J H, Dai W L . 2015. Spatial-temporal characteristics in urban morphology of major cities in China during 1990-2010. Economic Geography, 35(1): 44‒5. (in Chinese)

15
Ren B H . 2019. Study on spatial-temporal evolution of urban form and land use patterns in Kunming City. Diss., Kunming, China: Yunnan University of Finance and Economics. (in Chinese)

16
Shi F N, Liu S L, Sun Y X , et al. 2020. Ecological network construction of the heterogeneous agro-pastoral areas in the upper Yellow River Basin. Agriculture, Ecosystems & Environment, 302(10):107069. DOI: 10.101 6/j.agee.2020.107069.

17
Song Z J, Chen Y, Li Y . 2020. Comparative studies on evolutionary spatial multifractal mechanism for built-up lands in Zhengzhou from 1988 to 2015 with the characteristics of Beijing. Journal of Cleaner Production, 269(10):122451. DOI: 10.1016/j.jclepro.2020.122451.

18
Tan M H, Li Y Y . 2019. Spatial and temporal variation of cropland at the global level from 1992 to 2015. Journal of Resources and Ecology, 10(3): 235‒245.

19
Tian Y C, Ren Z Y . 2012. Land use change of Baota County, Yan’an City based on fractal dimension model. Journal of Arid Land Resources and Environment, 26(7): 184‒189. (in Chinese)

20
van Hees W W S . 1994. A fractal model of vegetation complexity in Alaska. Landscape Ecology, 9(4): 271‒278.

21
Wu M W, Zang C F, Fu J Y . 2020. Spatial and temporal variability characteristics and driving mechanism of land-use in Songliao River Basin from 1990 to 2015. Chinese Agricultural Science Bulletin, 36(32): 1‒9. (in Chinese)

22
Xiao D N, Burencang, Li X Z . 1997. Spatial ecology and landscape heterogeneity. Acta Ecologica Sinica, 17(5): 453‒461. (in Chinese)

23
Xie H L, Li X B . 2008. Spatial behavior characteristics of land use based on fractal theory: A case study in the East River Watershed, Jiangxi Province. Resources Science, 30(12): 1866‒1872. (in Chinese)

24
Xu B, Li Y F, Zhen Y . 2019. Changes of land use in rural-urban continuum of hilly regions―A case study on Sihe Township of Neijiang. Chinese Journal of Agricultural Resource and Regional Planning, 40(12): 24‒3. (in Chinese)

25
Xu J H, Ai N S, Jin J , et al. 2001. A fractal study on the mosaic structure of the landscape of Northwest China―Taking the drainage area of Heihe River as an example. Arid Zone Research, 18(1): 35‒39. (in Chinese)

26
Yu X J, Zhao Z Q, Yu L . 2019. Study on the complex features of Xiamen boundaries based on fractal theory. Journal of Natural Science of Heilongjiang University, 36(6): 738‒744. (in Chinese).

Options
Outlines

/

〈 〉