Changes of wetland landscape patterns in Dadu River catchment from 1985 to 2000, China

Abstract

Based on the interpretation and vector processing of remote sensing images in 1985 and 2000, the spatial changes of wetland landscape patterns in Dadu River catchment in the last two decades were studied using spatial analysis method. Supported by Apack software, the indices of wetland landscape pattern were calculated, and the information entropy (IE) was also introduced to show the changes of wetland landscape information. Results showed that wetland landscape in this region was characteristic of patch-corridor-matrix configuration and dominantly consisted of natural wetlands. Landscape patterns changed a little with low fragment and showed concentrated distribution with partial scattered distribution during the period from 1985 to 2000. The values of patch density and convergence index kept stable, and the values of diversity, evenness indices and IE showed a slight decrease, while dominance and fractal dimension indices were increased. All types of wetland landscapes had higher adjacency probabilities with grassland landscape in 1985 and 2000, and there was extremely weak hydrological link and large spatial gap among river, glacier, reservoir and pond wetlands due to low adjacency matrix values. Since their cumulative contribution exceeded 81% through the PCA analysis, the agriculture activities would be the main driving force to the landscape changes during the past 15 years.

Keywords

wetland landscape / spatial pattern / landscape indices / wetland management / Dadu River catchment

Introduction

Wetland, along with forest and ocean, is the important ecosystem in the world (Bai et al., 2005). Wetland values have been increasingly recognized, due to the important role in carbon and nutrient cycling, sediment accretion, pollution infiltration, and erosion controlling (Sahagian and Melack, 1998; Montalto et al., 2006; Qin and Mitsch, 2009). However, wetlands are also fragile ecosystems. O’Connell (2003) indicated that more than 50% of the world’s wetlands had been altered, degraded or lost through a wide range of human activities in the past 150 years, and only a small percentage of the original wetlands remained around the world after over two centuries of intensive development and urbanization. Thus the studies on the conservation and utilization of wetland become a worldwide issue.

Before 1970s, most wetland researchers had focused on the survey and utilization of wetland resources, as well as wetland protection and management (Elliot, 1968). Since 1980s, an increasing interest had occurred in ecological processes of wetlands, e.g., carbon and nitrogen cycling (Bennion, 1994), and ecological hydrological processes of wetland systems (Brown, 1989). To match the progress and needs of the researches, the quantification of landscape pattern and the interaction between landscape pattern and ecological processes, and the detection of landscape dynamics and functions became more essential (Tischendorf., 2001; Chen et al., 2008). Along with the development of some advanced tools, such as Remote Sensing, Global Position System, Geographic Information System, and some quantitative methods, i.e., spatial pattern indices (Turner and Gardner, 1991; Kong et al., 2007; Bai et al., 2008; Chen et al., 2008), which made the researches on wetland landscape patterns, structures and functions, and the dynamic changes more facile (Xiao et al., 1991; Thibault and Zipperer, 1994; Brazner et al., 2007; Zhang et al., 2008). Recently, most researches have focused on wetland landscape patterns in inland or coastal regions (Yue et al., 2003; Li et al., 2009), but little information is available on the changes of wetland landscapes in the southeast edge of the Qinghai-Tibet Plateau, which is one of potential regions influenced by the west line of South-to-North Water Diversion Project of China. Liu and Li (2006) and Canziani et al. (2006) also presented that the changes of watershed wetland landscape were greatly determined by dam construction, which had significant impacts on the longitudinal and transverse gradients and the internal structure of wetlands landscape within a watershed.

Dadu River catchment serves as one water source of the Yangtze River region. It plays an important role in maintaining the quantity and quality of water. Thus, more attention should be paid to the potential effects on wetland landscape before the operation of South-to-North Water Diversion Project. The objective of this study is to investigate the changes of wetland landscapes and their driving factors in the past 15 years in Dadu River catchment.

Materials and methods

Study area

With the total area of 77.4 × 10³ km², Dadu River catchment is one of important water supplying sources of the west line of South-to-North Water Diversion Project. It is located at the southeast edge of the Qinghai-Tibet Plateau and in the transition zone of the middle west of Sichuan Basin (31°3′–33°39′ N, 99°37′-102°43′ E). The upper reach of the Dadu River is located in the plateau frigid sub-humid zone. The middle reach is in the humid subtropical climate region with obvious dry and wet seasons, while the lower reach has the warm and humid subtropical climate without obvious seasonal change, with annual mean precipitation of 941 mm and annual mean runoff of 47 × 10⁹ m³. Meadow soil and limestone soil are two major soil types. Vegetation pattern shows vertical band spectrums along the increasing elevations (Wang et al., 2008).

Data sources

Topographic map, soil map, and vegetation map (scale 1∶100000, Aerial photographed in 1966, transferred in 1966, the first edition published in 1970; spatial resolution= 2.5 m) were chosen as references for the accuracy testing and validation. Landsat TM/ETM images in 1985 and 2000 (spatial resolution= 30 m) were provided by Institute of Geographic Sciences and Natural Resources Research of the Chinese Academy of Sciences, which were used to interpret the wetland spatial pattern data during the past 15 years. The precipitation, temperature, population, and agricultural data (fishery, cropland and irrigated area) were collected from the Sichuan Yearbook (Huang, 2000) to detect the main driving force of the changes of wetland landscape patterns.

Wetland landscape classification system

Various classification systems for wetland have been set up by different researchers (Tang and Huang, 2003). Generally these classification systems for wetlands based on the definition of wetland from the Ramsar Convention, and other classification systems for wetlands, as well as wetland vegetation characteristics of study area (Li et al., 2010). In this study, a third-grade classification was adopted by remote sensing classification of wetlands according to TM remote sensing data scale in two time series. The main landscape types of this region were listed in Table 1.

Data processing

Before remote sensing image interpretation, some extra data processing should be done, including fine geometric correction, image composition and enhancement. Then all spatial data were united to the Albers projection with two-standard line (Li et al., 2010).

To improve interpreting precision, human-machine interactive remote sensing interpretation (unsupervised classification and supervised classification methods) together with field survey were used to interpret the images in 1985 and 2000. Synthetic images were first created to templates by unsupervised classification, and then the researching samples were added to the templates. After that, all kinds of wetland and non-wetland templates were selected and trained by the ENVI ROI tools, and were tested and corrected according to the field investigation in June of 2008 during the interpretation until the satisfactory recognition of the templates could be gained (Li et al., 2010). Finally, the images were classified by means of maximum likelihood supervised classification (Yue et al., 2003). The wetland landscape maps of Dadu River catchment in 1985 and 2000 were illustrated in Fig. 1.

Wetland landscape pattern index

Landscape pattern index had been widely adopted to quantitatively describe spatial feature of landscape patterns at the patch, class and landscape levels (Thibault and Zipperer, 1994; Li et al., 2005). However, single index would not be adequate to interpret the information of spatial pattern, and some landscape pattern indices had the same or even contradictory ecological explanation (Schumaker, 1996). Therefore, those landscape indices which were easy to produce redundancy would be excluded (Riitters et al., 1995; Bai et al., 2008). Also, most of the indices were sensitive to certain scale-resolution, yet others not, therefore, more attention should be paid to the relationships between index values and ecological processes rather than the numbers of index (Li et al., 2005). Patch density (PD), convergence (RC), fractal dimension (FD), and internal habitat fragmentation (FI₁) indices were selected for the class level. Diversity (H), dominance (D), and evenness (E) indices were selected for the landscape level in this study due to the above reasons (O’Neill et al., 1988; Wang et al., 1996; Bai et al., 2008). All the indices were calculated using the APACK 2.0 software package (Mladenoff and Dezonia, 1997) with the grid file of 1000m × 1000 m cell siz

Adjacency matrix (AM)

The AM can be expressed as follows

(1)AM=\left[\begin{array}{llll}{A}_{\mathrm{11}}\hfill & A_{\mathrm{12}}\hfill & \cdots \hfill & A_{\mathrm{1}j}\hfill \\ A_{\mathrm{21}}\hfill & A_{\mathrm{22}}\hfill & \cdots \hfill & A_{\mathrm{2}j}\hfill \\ ⋮\hfill & ⋮\hfill & \hfill & ⋮\hfill \\ A_{i\mathrm{1}}\hfill & A_{i\mathrm{2}}\hfill & \cdots \hfill & A_{ij}\hfill \end{array}\right],

where AM represents the adjacency matrix probabilities between attribute classes. A_ij is the number of pixels in the raster map of category i that are adjacent to category j. The values of adjacency probabilities range from 0 to 100% and describe the proportional breakdown of neighbor cells (Johnson and Patil, 1998).

Spatial changes of landscape patches

Spatial changes of wetland landscape can be studied according to the centroid change of wetland landscapes and calculated as follows (Bai et al., 2008):

(2)X_t={\displaystyle \sum _{i=\mathrm{1}}^n(C_{ti}\times X_i)}/{\displaystyle \sum _{i=\mathrm{1}}^n{C}_{ti}},

(3)Y_t={\displaystyle \sum _{i=\mathrm{1}}^n(C_{ti}\times Y_i)}/{\displaystyle \sum _{i=\mathrm{1}}^n{C}_{ti}},

where, X_t and Y_t are the abscissa and ordinate of the centroid of wetland landscape in t period, respectively; X_i and Y_i are the abscissa and ordinate of the centroid of class i in the same period; C_ti denotes the area of each wetland patch in t period; {\displaystyle \sum _{\mathit{i}=1}^{\mathit{n}}{C}_{ti}} denotes the sum of C_ti.

Changes of information entropy (IE)

(4)\mathit{S}\mathit{T}=\left|{\displaystyle \frac{\mathit{I}{\mathit{E}}_{\mathit{i}+\mathrm{1}}}{{\mathit{N}}_{\mathit{i}+\mathrm{1}}}}-{\displaystyle \frac{\mathit{I}{\mathit{E}}_{\mathit{i}}}{{\mathit{N}}_{\mathit{i}}}}\right|\frac{1}{\mathrm{\Delta }\mathit{T}},

where, ST represents the changes of the IE during the exactly evolution periods- \mathrm{\Delta }T. \mathrm{\Delta }T denotes the time gap from i to i + 1 periods. ST∈[0,1). If ST = 0, it implies there are no changes in landscape structure. If ST = 1, it indicates the greater change in wetland landscape structure.

(5)\mathit{I}{\mathit{E}}_{\mathit{i}}=-\left({\displaystyle \sum _{\mathit{j}=\mathrm{1}}^{{\mathit{N}}_{\mathit{i}}}{\mathit{p}}_{\mathit{i}{\mathit{A}}_{\mathit{j}}}}\mathrm{l}\mathrm{n}{\mathit{p}}_{\mathit{i}{\mathit{A}}_{\mathit{j}}}+{\displaystyle \sum _{\mathit{j}=\mathrm{1}}^{{\mathit{N}}_{\mathit{i}}}{\mathit{q}}_{\mathit{i}{\mathit{A}}_{\mathit{j}}}}\mathrm{l}\mathrm{n}{\mathit{q}}_{\mathit{i}{\mathit{A}}_{\mathit{j}}}\right),

(6)I{E}_{i+\mathrm{1}}=-({\displaystyle \sum _{j=\mathrm{1}}^{N_{i+\mathrm{1}}}{p}_{i+\mathrm{1}{A}_j}}\ln p_{i+\mathrm{1}{A}_j}+{\displaystyle \sum _{j=\mathrm{1}}^{N_{i+\mathrm{1}}}{q}_{i+\mathit{1}{A}_j}}\mathrm{l}\mathrm{n}{q}_{i+\mathit{1}{\mathrm{A}}_j}),

where, IE_i and IE_i+1 are the values of IE in i and i + 1 period, respectively; N_i and N_i+1 represent the number of wetland types in i and i + 1 period; p_iAj and p_i+1Aj represent the area fractions of the wetland landscape, while q_iAj and q_i+1Aj represent the patch number fractions in i and i + 1 periods, respectively.

These equations are based on information theory originated from Shannon (1948), cross-entropy notion introduced by Kullback (1959) and Good (1963) (seen in You and Wood, 2005), and generalized entropy extended by Rènyi (1970) (seen in Carranza et al., 2007), but there are some difference in this study, the number proportion of the patches (q_iAj) is also included in IE_i besides the area proportion (p_iAj), and the time gap \mathrm{\Delta }T is considered in ST.

These equations reflect the information of the spatial geometrical permutation and combination of whole the landscape. This is just as distribution law in thermodynamics (Jaynes, 1979), which is defined as a measure of the “disorder” of molecules in a system.

Results and discussion

Spatial distribution characteristics of wetland landscapes

Structure of wetland landscape

During the period from 1985 to 2000, the wetland landscape dominantly consisted of natural wetlands with the proportion more than 24 times as much as that of constructed wetland (Table 2), in which marsh wetland was the dominated wetland landscape covering more than 74% of the total wetland area. According to Forman (1983) theory, wetland landscape pattern of this region was typically characteristic of patch-corridor-matrix configuration, with marsh wetland serving as the matrix, river wetlands as the corridor, and both lake wetland and glacier wetland as the dominant patches.

The total number of patches of natural wetlands was about six times as much as that of constructed wetland, in which the number of patches of marsh and lake wetlands was seven times more than that of other wetland types. The average patch area of natural wetland landscape was larger than constructed wetland landscape during the study period. The maximum area of patches followed the order of marsh wetland>river wetland>glacier wetland>lake wetland>reservoir and pond wetland, while the average patch area was in the order of glacier wetland>river wetland>marsh wetland>reservoir and pond wetland>lake wetland (Table 2).

Wetland landscape pattern index

Diversity index and dominance index changed a little and kept higher value from1985 to 2000, which indicated that there were big differences among the proportions of different wetland landscape types and each proportion generally maintained constant. Wetland landscape kept stably high convergence index (RC>0.99) and low patch density (PD = 0.014) in both years (Table 3), which implied the concentrated distribution of wetland landscape with low fragmentation level in the past 15 years. The wetland patches showed complicated geometric shapes with weak self-similarity with the value of fractal dimension index above 1.42 in the past two decades. This was mainly associated with few disturbances of human activities and the establishment of wetland nature reserve after 1990s in this region (An et al., 2008; Gibbes et al., 2009).

Among the five landscape types, a scattered distribution was observed both in lake wetland, and reservoir and pond wetland landscapes due to higher values of PD and FI₁ indices, while river, marsh, and glacier wetland landscapes showed lower values of both indices. Compared to river wetland, other wetland landscapes, especially the reservoir and pond wetland landscape had simpler and more regular geometric shape (Fig.2) due to lower FD value. Wang etal. (2009) also presented that intense disturbance made the shape of the patches much simpler or more regular.

Because most landscape indices originated from statistics or mathematics, and indicated the landscape geometric characteristics of wetlands at patch and landscape levels, yet they could not present their spatial and functional characteristics observed from aerial photos (Lin et al., 2008). Chen et al. (2008) also pointed out that some landscape indices would not change, whereas the effects of landscape pattern on ecological processes probably changed greatly. In this study, the PD and RC didn’t change during the period from 1985 to 2000, but it was difficult to reveal the stable ecological processes in the past 15 years, because landscape indices were mathematical constructs that had no inherent ecological meaning (Xiao et al., 1991; Li and Reynolds,1993). Keddy (1992) and Brazner et al. (2007) further suggested that functional indicators were those based on species traits rather than taxonomy, because species distribution referred to many factors, such as climate, wetland sources, and vegetation patterns.

Adjacency matrix probabilities between wetland and non-wetland landscapes

The adjacency matrix probabilities were calculated by Eq. (1), and these metrics were robust in the sense of quantifying both composition and configuration of landscape pattern (Li and Reynolds, 1993). During the study period, all types of wetland landscape had higher adjacency probabilities with grasslands (Tables 4 and 5), which demonstrated that there were close correlations between wetland and grassland since grassland landscape was the landscape matrix in this region. However, there were much lower adjacency probabilities between all types of wetland landscapes, which suggested that there were weak correlations and interactions among all types of wetland landscapes in this region.

Except for AM value between lake wetland and marsh wetland, the AM values between other wetland types were 0 (Tables 4 and 5), which indicated there was an extremely weak hydrological link and large spatial gap among river, glacier, reservoir and pond wetlands. Additionally, lake wetland had higher adjacency probabilities with woodland, grassland, human habitant and dry-land than other wetland types due to much more scattered distribution, which was supported by the results of landscape pattern indices (Fig.2). Other than an increasing AM value between river wetland and grassland, that a decreasing AM value occurred between other wetlands and grasslands from 1985 to 2000. This implied wetland degradation because there was increasing spatial gaps between these wetlands, and the whole wetland landscape system became frailer than past (Cabezas et al., 2008). Herkert et al. (2003) also reported that all these changes would lead to a bad consequence on biodiversity, water and carbon fluxes at local or regional levels.

Spatial and temporal changes in wetland landscape pattern

Landscape pattern changes can be distinguished into land-use and land-cover changes, and landscape position changes. Modification and conversion are two important processes of the land-use and land-cover changes. Modification presents land cover types do not change in despite of changes in some properties, while conversion means one land cover type is converted to another one (Turner and Meyer, 1998; Yue et al., 2003). Based on the theory stated above, we defined the modification of wetland landscapes as the transformation between wetland types, and defined the conversion of wetland landscapes as the transformation between wetland and non-wetland landscapes. The wetland landscape position changes here can be described by centroid changes of wetland patches. The centroid indicated area central tendency, especially the weight meaned the center of spatial distribution (Yue et al., 2003).

Transformation between wetland and non-wetland landscapes

Table 6 showed that there was no obvious modification between wetland types due to weak link among them, which is in agreement with the aforementioned results of adjacent probability. Lower adjacency matrix probabilities among all types of wetland landscapes indicated that weak correlations and interactions showed among these wetland landscape types in this region, so it was difficult to modify one type to another one. The conversion of wetland and non-wetland landscapes just occurred between lake wetland and woodland or grassland landscapes, as well as between river and marsh wetland and grassland landscapes. No conversions occurred between constructed wetlands and non-wetlands (Fig. 3).

As Fig. 3 shown, the area of lake wetland generally reduced, wherein about 88.10 hm² lake wetlands were converted to the grassland. Less than 1 hm² lake wetlands were converted to woodlands. In contrast, the areas of both river and marsh wetland were increased from 1985 to 2000. In total of 155.94 hm² grasslands were converted to river wetlands. There was also a net increase by 1027.97hm² for marsh wetland during the study period due to the conversion from grassland. This was related to the fact that the buffer zones of river wetland are located in the alpine zone, and marsh wetlands are located in the upper reach of Dadu River catchment. During the period from 1985 to 2000, approximately 1183 hm² non-wetlands were converted to wetland landscapes, while only 88.16hm² wetlands were lost during the study period. The total wetland area increased by 1095.79 hm², accounting for about 6.65% of the wetland area in 2000 (Table 3 and Table7), because wetlands were well conserved in this region in the past two decades.

Changes in the centroids of wetland patches

Spatial changes in wetland landscape were described according to the centroid changes of wetland landscapes. The centroid of each wetland patch was calculated and transferred to coverage files using the centroid labels order of Arc/Info, and their latitudes and longitudes could be obtained by transferring the projection system to geographic coordinates. The centroid changes of wetland landscape during the period from 1985 to 2000 were illustrated in Fig. 4.

Owing to the effects of both human activities and natural environmental evolution (Yue et al., 2003), the centroid of marsh wetland moved 1.16º in the south direction and then moved 0.08º in the west, thus moved 127.9 km in the southwest direction. However, the centroids of river wetland, lake wetland, glacier wetland, and reservoir and pond wetland landscapes did not move in the past 15 years. This did not indicate no changes in the total areas of river, lake, glacier and reservoir and pond wetland landscapes (e.g., the area of river wetland was showed an increase by155.94 km², while a decrease by 88.12 km² for lake wetland; Table 2 and Fig. 3). In contrast, the contradiction further suggested that the changing direction of marsh wetland centroid was unidirectional, other types of wetland were polydirectional.

Changes of IE

IE₂₀₀₀, IE₁₉₈₅, and ST were calculated by the Eqs. (4)–(6). The results were shown as follows:

IE₁₉₈₅ = 2.08; IE₂₀₀₀ = 2.06; ST = 0.00029.

There was a little decrease in the value of IE, which illustrated that the information of the wetland landscapes was mildly reduced. With and King (1999) concluded that gap structure of landscape was a more important determinant of dispersal than patch structure on dispersal success of fractal landscape.

The value of ST illustrates the spatial geometrical permutation and combination of landscape. This implies the integrated stability of wetland landscape. During the study period, the extremely small values of ST indicated that the structure information of wetland landscape (the area proportion and patch proportion of each wetland type) changed a little and slowly (Table 8). The tectonic disciplinarian and regulation of the whole wetland landscape kept stable, even though the area and number of patches, some landscape pattern indices, and AM values changed a little (Tables 2, 4, 5, and 8).

Driving factors of wetland landscape changes

Human activities and natural factors are two main driving factors of landscape changes and wetland degradation (An et al., 2008). An integration of natural factors and human activities in the explanation of land-use and land-cover dynamics remains as an important research task (Sluiter and de Jong, 2007).

Human activity is the major driving force of shaping land-use and land-cover pattern, although the underlying physical structure of landscape may constrain land-use and land-cover changes (Pan et al., 1999; Serraa et al., 2008). However, changes in human activities often produce changes in spatial landscape heterogeneity due to the change in old ecosystem and the creation of new one (Carranza et al., 2007), thus the effects were particularly true in wetlands (Alvarez-Rogel et al., 2007). A variety of human activities including clearing for agriculture, harvesting of timber and other natural resources, and urbanization can affect land cover patterns and composition, which in turn can also alter the functions of ecosystem (Gibbes et al., 2009; Wang and Zhang, 2002).

Owing to the limitation of the data (the integrity and availability of the data), the following data were selected in this study. Human activity data mainly contained urbanization factors-agriculture population, the ratio of agriculture population to total population; agriculture factors-forestry production, animal husbandry production, fishery production, the area of cropland and irrigation, the number of livestock, other factors—total population, GDP; and the natural factors mainly included precipitation and temperature. Detail data were listed in Table 9.

The main influencing factors were identified using principal component analysis (PCA). Based on the eigenvalues (eigenvalue>1), three main principal components (PCs) explains 94.56% of the total variance (Table10). The first PC1, explaining 59.3% of the total variance, was strongly and positively related to total population, agriculture population, forestry production, animal husbandry production, fishery production, the number of livestock, and the area of cropland (Table 10). The second PC2 explained 21.71% of the total variance, and showed higher positive factor loadings on GDP and the irrigated area and negative factor loadings on the area of cropland and air temperature. The third PC3 explaining 12.55% of the total variance, showed higher positive factor loadings on precipitation and air temperature.

The first PC1 and the second PC2 were named as agriculture factor and the third principal component was climate factor. In view of the cumulative contribution exceeded 81%, the agriculture factor would be the dominated factor influencing the wetland landscape changes in Dadu River catchment. However, owing to the limits of geographical conditions, the development of the urbanization was slowly, and the effects of the climate was very weak (Tables 10 and 11). Most researchers also presented that agricultural drainage and agricultural management were important threats to wetland ecosystems (Hodge and McNally, 2000; Hansson et al., 2005; Bai et al., 2008; Niedermeier and Robinson, 2009).

According to climate data, no obvious changes of annual mean precipitation were observed, but annual mean air temperature showed a slight increase in this region during the period from 1985 to 2000 (Table 9), which had weak effects on wetland landscapes and contributed to keeping the whole wetland landscape’s structure stable during the study period. Air temperature could result in the melting of the ice and snow on the mountain to some extent, however, glacier wetland had invisible change due to the remote sensing images with resolution of 30m, which would contribute to water supply to the rivers and thus the increase in river wetlands. The marsh wetland also showed a slight increase during the study period. Additionally, the lake wetland area showed a slight decrease due to the conversions to other land uses (i.e., grassland) (Fig. 3). Because the driving factors were responsible for the changes represent complex characteristics, further study should be done on the relationship between wetland area and climatic factors in different regions.

Table 9 showed the development of fishery in the past years from 1989 to 1999. The increasing rate of the fishery was low, and the development of fishery might be related to the increase in investment and utilization, so the area and number of the reservoir and pond wetland did not show an obvious increase during the study period. The conversion from farmland to forest and grassland had been implemented since 1989 (Table 9), thus the areas of the cropland and irrigated farmland greatly decreased from 1989 to 1999. This led to an increase in the area of woodland and grassland landscape. The proportions of agricultural population to total population also dropped from approximately 86% in 1989 to 82% in 1999 (Table 9), indicating that the effects of agricultural tillage on wetland landscapes became weaker from 1985 to 2000. Some grassland landscape had also been converted to wetland landscapes due to the establishment of wetland reserve since 1992, which resulted in an increase in wetland area of 1095.79hm². Almost all the conversions occurred between grasslands and wetlands, due to the highest adjacency probabilities between them during the period from 1985 to 2000 (Tables4 and 5).

It is necessary to find a proper way to alleviate all these potential influences of human activities and hydrological engineering on the wetland ecosystem. Chen et al. (2008) proposed that it may provide an access to this if the Patch-Corridor-Matrix theory could be applied to solve practical problems. Marsh wetland landscape as the matrix with the highest proportion of the total wetland area (>74%) should be protected undoubtedly, especially the large patches and the better connected patches (lower PD and FI₁ indices of marsh wetland suggested the ecological conservation significance of this type of wetland), river wetlands as the corridor can be prolonged alongside the main axis and would build much more hydrological links among wetland patches, thus increase the exchange, discharge, recharge of water, and then ameliorate the degradation of wetland ecosystem (Mikhailov et al., 2003; Huang et al., 2005; Valdemoro et al., 2007; Cabezas et al., 2008).

Conclusions

The study area had a typical patch-corridor-matrix configuration, in which marsh wetland was the landscape matrix, river wetland as the corridor, and lake and glacier wetlands as the main patches. Wetland landscape structure kept stable. Marsh wetland was dominant wetland landscape during the period from 1985 to 2000. The wetland area was increased by 1095.79 hm² in the Dadu River catchment in the past 15 years. Agricultural activity was the dominate factor which contributed to the changes of wetland landscape pattern. The findings of this study can provide a reliable scientific support for the ecological hydrological control function of wetland ecosystem on Water Diversion Projects, as well as to give the theoretical guidance to the wetland protection and management in the diversion and flooded regions during the operation of the hydrological engineering. Additionally, this study also provided an approach to investigate the changes of wetland landscape pattern.

References

[1] Álvarez-Rogel J, Jiménez-Cárceles F J, Roca M J, Ortiz R (2007). Changes in soils and vegetation in a Mediterranean coastal salt marsh impacted by human activities. Estuar Coast Shelf Sci, 73(3–4): 510–526

[2] An N, Gao N, Liu C (2008). Wetland degradation in China: causes, evaluation, and protection measures. Chin J Ecol, 27(5): 821–828 (in Chinese)

[3] Bai J H, Ouyang H, Yang Z F, Cui B S, Cui L J, Wang Q G (2005). Changes in wetland landscape pattern: a review. Progr Geogr, 24(4): 36–45 (in Chinese)

[4] Bai J H, Ouyang H, Cui B S, Wang Q G, Chen H (2008). Changes in landscape pattern of alpine wetlands on the Zoige Plateau in the past four decades. Acta Ecol Sin, 28(5): 2245–2252

[5] Bennion H (1994). A diatom phosphorus transfer function for shallow, eutrophic ponds in southeast England. Hydrobiologia, 275/276: 391–410

[6] Brazner J C, Danz N P, Niemi G J, Regal R R, Trebitz A S, Howe R W, Hanowski J M, Johnson L B, Ciborowski J J H, Johnston C A, Reavie E D, Brady V J, Sgro G V (2007). Evaluation of geographic, geomorphic and human influences on Great Lakes wetland indicators: a multi-assemblage approach. Ecol Indic, 7(3): 610–635

[7] Brown M T(1989). A simulation model of hydrology and nutrient dynamics in wetlands. Comput Environ Urban Syst, 12(4): 221–237

[8] Cabezas Á, González E, Gallardo B, García M, González M, Comín F A (2008). Effects of hydrological connectivity on the substrate and understory structure of riparian wetlands in the Middle Ebro River (NE Spain): implications for restoration and management. Aquat Sci, 70(4): 361–376

[9] Canziani G A, Ferrati R M, Rossi C, Ruiz-Moreno D (2006). The influence of climate and dam construction on the Ibera wetlands, Argentina. Reg Environ Change, 6(4): 181–191

[10] Carranza M L, Acosta A C, Ricotta C (2007). Analyzing landscape diversity in time: the use of Rènyi’s generalized entropy function. Ecol Indic, 7(3): 505–510

[11] Chen L, Liu Y, Lü Y, Feng X, Fu B (2008). Pattern analysis in landscape ecology: progress challenges and outlook. Acta Ecol Sin, 28(11): 5521–5531

[12] Elliot S H (1968). IUCN ecology commission’s technical meeting in Turkey on wetlands conservation in Western Asia, October 1967. Biol Conserv, 1(1): 97–98

[13] Forman R T T (1983). Corridors in a landscape: their eco-logical structure and function. Ecology, 2: 375–387

[14] Gibbes C, Southworth J, Keys E (2009). Wetland conservation: change and fragmentation in Trinidad’s protected areas. Geoforum, 40(1): 91–104

[15] Good I J (1963). Maximum entropy for hypothesis formulation, especially for multidimensional contingency tables. Ann Math Stat, 34(3): 911–934

[16] Hansson L A, Brönmark C, Anders N P, Åbjörnsson K (2005). Conflicting demands on wetland ecosystem services: nutrient retention, biodiversity or both? Freshw Biol, 50(4): 705–714

[17] Herkert J R, Reinking D L, Wiedenfeld D A, Winter M, Zimmerman J L, Jensen W E, Finck E J, Koford R R, Wolfe D H, Sherrod S K, Jenkins M A, Faaborg J, Robinson S K (2003). Effects of prairie fragmentation on the nest success of breeding birds in the mid-continental United States. Conserv Biol, 17(2): 587–594

[18] Hodge I, McNally S (2000). Wetland restoration, collective action and the role of water management institutions. Ecol Econ, 35(1): 107–118

[19] Huang H J, Li F, Pang J Z (2005). Interaction Research of Land–Ocean Between Yellow Sea and Bohai. Beijing: Science Press, 61–64 (in Chinese)

[20] Huang G Q (2000). Sichuan Statistical Yearbook. Beijing: Chinese Statistical Press, 101–196 (in Chinese)

[21] Jaynes E T (1979). Where do we stand on maximum entropy? In: Levine R D, Tribus M, eds. The Maximum Entropy Formalism. Cambridge: MIT Press, 1–105

[22] Johnson G D, Patil G P (1998). Quantitative multiresolution characterization of landscape patterns for assessing the status of ecosystem health in watershed management areas. EcoHealth, 4(3): 177–187

[23] Keddy P A (1992). A pragmatic approach to functional ecology. Funct Ecol, 6(6): 621–626

[24] Kong C F, Xu K, Wu C L (2007). Research on the landscape change of Yeya Lake wetland based on remote sensing fusion. In: Conference on Remotely Sensed Data and Information. 2007, May, 25th–27th, Nanjing, China, 6752(2): 277–786

[25] Kullback J (1959). Information Theory and Statistics. New York: John Wiley & Sons, 353–393

[26] Li A, Deng W, Kong B, Song M, Feng W, Lu X, Lei G, Bai J (2010). A comparative analysis on spatial patterns and processes of three typical wetland ecosystems in 3H area, China. Procedia Environmental Sciences, 2: 315–332

[27] Li H, Reynolds J F (1993). A new contagion index to quantify spatial patterns of landscapes. Landscape Ecol, 8(3): 155–162

[28] Li S, Wang G, Deng W, Hu Y, Hu W (2009). Influence of hydrology process on wetland landscape pattern: a case study in the Yellow River Delta. Ecol Eng, 35(12): 1719–1726

[29] Li X, Jongman R H G, Hu Y, Bu R, Harms B, Bregt A K, He H S (2005). Relationship between landscape structure metrics and wetland nutrient retention function: a case study of Liaohe Delta, China. Ecol Indic, 5(4): 339–349

[30] Lin M L, Cao Y, Wang S (2008). Limitations of landscape pattern analysis based on landscape indices: a case study of Lizejian wetland in Yilan of Taiwan Province, China. Chin J Appl Ecol, 19(1): 139–143

[31] Liu H Y, Li Z F (2006). Spatial gradients of wetland landscape and their influential factors in watershed. Acta Ecol Sin, 26(1): 213–220

[32] Mikhailov V N, Kravtsova V I, Magritskii D V (2003). Hydrological and morphological processes in the Kura river delta. Water Res, 30(5): 495–508

[33] Mladenoff D J, Dezonia B (1997). APACK 2.0 User’s Guide. Department of Forest Ecology and Management, University of Wisconsin-Madison, Madison, USA

[34] Montalto F A, Steenhuis T S, Parlange J Y (2006). The hydrology of Piermont Marsh, a reference for tidal marsh restoration in the Hudson river estuary, New York. J Hydrol (Amst), 316(1–4): 108–128

[35] Niedermeier A, Robinson J S (2009). Phosphorus dynamics in the ditch system of a restored peat wetland. Agric Ecosyst Environ, 131 (3–4): 161–169

[36] O’Connell M J (2003). Detecting, measuring and reversing changes to wetlands. Wetlands Ecol Manage, 11(6): 397–401

[37] O’Neill R V, Krummel J R, Gardner R H, Sugihara G, Jackson B, DeAngelist D L, Milne B T, Turner M G, Zygmunt B, Christensen S W, Dale V H, Graham R L (1988). Indices of landscape pattern. Landsc Ecol, 1(3): 153–162

[38] Pan D, Domon G, de Blois S, Bouchard A (1999). Temporal (1958–1993) and spatial patterns of land use changes in Haut-Saint-Laurent (Quebec, Canada) and their relation to landscape physical attributes. Landsc Ecol, 14(1): 35–52

[39] Qin P, Mitsch W J (2009). Wetland restoration and ecological engineering: international conference of wetland restoration and ecological engineering. Ecol Eng, 35(4): 437–441

[40] Rènyi A (1970). Probability Theory. Amsterdam: North Holland Publishing Company, 646–660

[41] Riitters K H, O’Neill R V, Hunsaker C T, Wickham J D, Yankee D H, Timmins S P, Jones K B, Jackson B L (1995). A factor analysis of landscape pattern and structure metrics. Landsc Ecol, 10(1): 23–39

[42] Sahagian D, Melack J (1998). Global Wetland Distribution and Functional Characterization: Trace Gases and the Hydrologic Cycle. IGBP GAIM Report No.2, 40

[43] Schumaker N H (1996). Using landscape indices to predict habitat connectivity. Ecology, 77(4): 1210–1225

[44] Serraa P, Pons X, Saur D (2008). Land-cover and land-use change in a Mediterranean landscape: a spatial analysis of driving forces integrating biophysical and human factors. Appl Geogr, 28(3): 189–209

[45] Shannon C E (1948). A mathematical theory of communication. Bell Syst Tech J, 27: 379–423, 623–656

[46] Sluiter R, de Jong S M (2007). Spatial patterns of Mediterranean land abandonment and related land cover transitions. Landscape Ecol, 22(4): 559–576

[47] Tang X, Huang G (2003). Study on classification system for wetland types in China. For Res, 16: 531–539 (in Chinese)

[48] Thibault P A, Zipperer W C (1994). Temporal changes of wetlands within an urbanizing agricultural landscape. Landsc Urban Plan, 28(2–3): 245–251

[49] Tischendorf L (2001). Can landscape indices predict ecological processes consistently? Landscape Ecol, 16(3): 235–254

[50] Turner B L II, Meyer W B (1998). Global land-use and land cover change: an overview. In: Meyer W B, Turner B L II, eds. Change in Land Use and Land Cover: A Global Perspective. Cambridge: Cambridge University Press, 1–12

[51] Turner M G, Gardner R H (1991). Quantitative Methods in Landscape Ecology. New York: Springer-Verlag, 417–442

[52] Valdemoro H I, Sánchez-Arcilla A, Jiménez J A (2007). Coastal dynamics and wetlands stability. The Ebro delta case. Hydrobiologia, 577(1): 17–29

[53] Wang G, Zhang Y (2002). Impacts of reservoir project on hydrologial and ecologial environment of Xianghai wetlands. Resources Science, 24(3): 26–30 (in Chinese)

[54] Wang H, Yan D H, Qin D Y (2008). Water Source Area of West Line of South-to-North Water Diversion Project–Water Cycle Simulation and Quantitative Assessment of Water Resources. Zhengzhou: Water Conservancy of Yellow River Press, 28–44 (in Chinese)

[55] Wang X L, Bu R, Hu Y M, Xiao I (1996). Analysis on landscape fragment of Liaohe Delta wetlands. Chin J Appl Eco, 7(3): 299–304

[56] Wang Z, Song K, Zhang B, Liu D, Ren C, Luo L, Yang T, Huang N, Hu L, Yang H, Liu Z (2009). Shrinkage and fragmentation of grasslands in the West Songnen Plain, China. Agric Ecosyst Environ, 129(1–3): 315–324

[57] With K A, King A W (1999). Dispersal success on fractal landscapes: a consequence of lacunarity thresholds. Landsc Ecol, 14(1): 73–82

[58] Xiao D N, Zhao Y, Sun Z W (1991). Changes in spatial pattern of west suburb landscape in Shenyang City. In: Xiao D N, ed. Landscape Ecology: Theory, Methods and Applications. Beijing: Chinese Forestry Press, 186–195 (in Chinese)

[59] You L, Wood S (2005). Assessing the spatial distribution of crop areas using a cross-entropy method. Int J Appl Earth Observ Geoinform, 7(4): 310–323

[60] Yue T X, Liu J Y, Jørgensen S E, Ye Q H (2003). Landscape change detection of the newly created wetland in Yellow River Delta. Ecol Model, 164(1): 21–31

[61] Zhang X L, Cai H S, Ding S T, Huang S G (2008). Analysis on the landscape pattern changes of Poyang Lake wetlands and their driving force. J Anhui Agri Sci, 36(36): 16065–16070, 16078 (in Chinese)

RIGHTS & PERMISSIONS

Higher Education Press and Springer-Verlag Berlin Heidelberg