Monitoring Wetland Landscape Evolution Using Landsat Time-Series Data: A Case Study of the Nantong Coast, China

: Coastal wetlands not only have rich biodiversity and high productivity but also provide important ecological services. The monitoring of landscape changes can provide important support for the sustainable development of coastal zones. Landsat images from 1986 to 2017 were used to interpret the types of coastal wetlands in Nantong. A single dynamic degree and multiple landscape indices were calculated to analyze the rate of change and characteristics of each wetland type. The results demonstrate the following: (1) A Nantong wetland type system was established, which was divided into three major categories and eleven subcategories. (2) In general, natural wetlands, such as thatched and Suaeda salsa marshes, were extremely reduced, while artiﬁcial wetlands and non-wetlands with high human activity, such as breeding ponds, farmland, and construction land, increased signiﬁcantly. (3) In the past 30 years, due to the inﬂuence of environmental pressures, such as population growth, land demand, and economic development, the major inﬂuencing factors of local landscape change have shifted from natural geographical factors to human activities and economic as well as social factors. Remote sensing wetland interpretation can be very helpful in monitoring the dynamic changes in coastal wetlands and can provide scientiﬁc support for the sustainable management of coastal zones.


Introduction
Coastal systems and wetlands provide an important portion of global ecosystem services [1]. Coastal wetlands are located in the staggered transition zone between terrestrial ecosystems and marine ecosystems, and have important ecological, economic, and scientific value (valued as being up to USD 194,000 ha −1 yr −1 ) [2]. Coastal wetlands not only have rich biodiversity and high productivity but also provide important services that cannot be replaced by other ecosystems, such as carbon sequestration [3], environmental purification [4], climate regulation [5], flood resistance [6], and biodiversity conservation [7]. However, they are also vulnerable areas that are being rapidly lost under the dual pressures of rising sea levels and high-intensity human activity [8][9][10]. Previous studies have shown that 16.02% of tidal flats were lost from 1984 to 2016, mainly due to human reclamation activities [11]. Human activities have greatly changed the landscape patterns and land-use types of coastal wetlands [12,13]. The coastal wetlands in Jiangsu Province are notable in the Asian continental margin as having the most typical tidal flats developed offshore. However, with rapid economic development, terrestrial areas have notably increased, and large areas of coastal wetlands have decreased [14,15]. In order to utilize coastal wetland resources more rationally, it is necessary to explore management strategies of the sustainable utilization of coastal wetlands by tracking the history of these coastal wetlands.

Dataset
In order to investigate the spatial and temporal distribution characteristics of the Nantong coastal wetlands, we collected the satellite image data from the Landsat-5 TM,  The coastal wetlands in Nantong mainly include natural wetlands such as thatch, Suaeda salsa, Spartina, and bare tidal flats, as well as land-use types with a high intensity of human activities, such as farmland, breeding ponds, and construction land [42,43]. The research scope is the coastal area of Nantong. The land area was bounded by the coastline in 1980, and the sea area is about −6 m deep, shown as a black dotted line (Figure 1). This is due to the boundary of wetlands in marine water areas, which was defined in the 'Convention on Wetlands'.

Dataset
In order to investigate the spatial and temporal distribution characteristics of the Nantong coastal wetlands, we collected the satellite image data from the Landsat-5 TM, Landsat-7 ETM+, and Landsat-8 OLI images of the medium-resolution U.S. Landsat series shown in Figure 2, and recorded Spartina, Suaeda salsa, reed, agricultural land, breeding ponds, salt pans, river reservoirs, construction lands, reclamation areas, etc. in the field survey for the verification of remote sensing identification and classification. The time range of the images was from 1986 to 2017, and the details are shown in Table 1. The tide gauge data were obtained from the Yangkougang tide level station in the sea area of the RSRs ( Figure 1). The historical data before 2012 were calculated by a harmonic analysis [44]. All of the remote sensing images have undergone preprocessing, such as radiometric correction and geometric correction, and the error of geometric correction was controlled within one pixel. All of the images were projected onto 'WGS_1984_UTM_Zone_51N' datum planes.  According to the 'Convention on Wetlands,' China's wetland classification system, combined with the distribution of the Nantong coastal wetlands and the integrity of the ecosystem, a classification system of the Nantong coastal wetlands was established (Table  2) [45]. The system is generally divided into natural wetlands, artificial wetlands, and non-  According to the 'Convention on Wetlands,' China's wetland classification system, combined with the distribution of the Nantong coastal wetlands and the integrity of the ecosystem, a classification system of the Nantong coastal wetlands was established ( Table 2) [45]. The system is generally divided into natural wetlands, artificial wetlands, and non-wetlands, including 11 categories: Spartina salt marsh, Suaeda salsa salt marsh, reed marsh, thatched marsh, bare mudflat and offshore, river bank, breeding pond, salt pan, construction land, farmland and woodland, and others (abandoned land, reclamation of unused land, etc.).

Coastal Wetlands Interpretation
Through field surveys and remote sensing image comparisons, the coastal wetlands were interpreted to determine the interpretation signs of various landscape types and land-use types. For Swir, near-infrared, and red bands composite images of the Landsat images, the reed marshes are bright green and irregular in shape, and are mostly distributed along the supratidal zone outside the estuary; the thatched marsh area is yellow-green and irregular in shape; the vegetation of the Suaeda salsa is short, with dots of gray-red mixed with green spots; outside the Suaeda salsa is a dark green or brown Spartina salt marsh; and bare mudflats and offshore are gray-pink or dark gray due to differences in moisture content. Breeding ponds and salt pans are generally dark blue, and the distribution information of large-scale salt pans is clearly distinguishable from that of breeding ponds; the shape of farmland is generally regular, mostly grass green; woodland is green, but most of the coastal embankments are distributed in strips; and the construction land is generally irregular in shape, with red mottled spots.
According to the color tone, shape, texture, and spatial relationship of remote sensing images, as well as field survey photos, based on expert knowledge, the interpretation marks of Nantong coastal wetland types were determined (Table 3). A man-computer interactive method, combining supervised classification and artificial visual interpretation, was used to identify the wetland types via ENVI 5.2 software and ArcGIS 10.4 software. Two groups of training samples were established according to the marks and field observation points. One group was used for supervised classification by the method of maximum likelihood, and the other one was used to verify classification accuracy. The Kappa coefficients were all above 0.85, and the interpretation accuracy reached 90%.

. Landscape Change Analysis
The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension

. Landscape Change Analysis
The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension

. Landscape Change Analysis
The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017 In the formula, K is the dynamic degree of a wetland landscape; Ua and Ub are a wetland area at the initial stage and the end stage, respectively; and T is the interval years.
Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension

Landscape Change Analysis
The most direct manifestation of landscape pattern change is land-use change. Changes in land use lead to changes in the quantity and spatial combination of land areas, which, in turn, lead to changes in landscape patterns, thus affecting the matrix and structure of landscape types in a region [46]. Two methods were used for landscape change analysis.
A single landscape dynamic model (Formula (1)) was used to analyze the degree of change in the landscape area of the study area in different time periods. This study has four time periods : 1986-1993, 1993-2002, 2002-2008, and 2008-2017.
In the formula, K is the dynamic degree of a wetland landscape; U a and U b are a wetland area at the initial stage and the end stage, respectively; and T is the interval years. Landscape metrics or indices have been commonly used for analyzing land-use dynamics and urban growth processes [47][48][49]. Landscape indices, such as the number of patches, NP; patch density, PD; largest patch index, LPI; AREA_MN; fractal dimension index, FRAC_MN; contagion index, CONTAG; Shannon's diversity index, SHDI; Shannon's evenness index, SHEI; and the aggregation index, AI, were calculated by Fragstats 4.2 software.

Spatial Distribution of Wetland Types
The images of five periods in the coastal area of Nantong, Jiangsu were interpreted, and the classification results are shown in Figure 3. The results show that the composition of the entire Nantong coastal wetland landscape has changed significantly in the 30 years from 1986 to 2017 (Figure 4). The total area of coastal wetlands continued to decrease from 3555.25 km 2 in 1986 to 3328.55 km 2 in 2017. Among them, the natural wetland area decreased from 3551.91 km 2 in 1986 to 3179.08 km 2 in 2017. A total of 10.5% of natural wetlands have been lost over the past 30 years. The area of natural wetlands lost during 2008-2017 was significantly larger than that during 1986-1993, 1993-2002, and 2002-2008, and the loss rate of natural wetlands was also the largest. The area of artificial wetlands and non-wetland with a high intensity of human activities, such as breeding ponds, farmland, and construction land, increased significantly.
In natural wetlands, the salt marsh of Suaeda salsa decreased, while the salt marsh of Spartina increased, and the change in the reed and thatch was not obvious. Among them, the reed marsh first increased and then decreased from 17.36 km 2 in 1986, and increased again in recent years, reaching 58.59 km 2 in 2017, which was related to artificial planting and growth in the reclamation area. Thatched marsh has decreased significantly in recent years. The area of Suaeda salsa decreased from 77.21 km 2 in 1986 to 3.9 km 2 in 2017, with the greatest degree of degradation. In order to protect the coast and reduce coastal erosion, since the 1960s, Spartina has been introduced in Jiangsu, China, to promote siltation. The results in 1986 showed that the area of Spartina salt marshes reached 45.33 km 2 and has maintained a high growth range since then. However, due to the 1990s, the large-scale reclamation has reduced the area of Spartina to some extent, about 37.66 km 2 in 2002, but it still maintained growth due to its good environmental adaptability, and the area reached 59.14 km 2 in 2008. Under the background of large-scale reclamation, the area of Spartina salt marshes decreased, but there was still 47.11 km 2 in 2017. In a word, the main change characteristic of the Nantong coastal wetlands is that the natural wetlands represented by thatched grass and Suaeda salsa were greatly reduced.
Most of the artificial wetlands and farmland, such as breeding ponds and salt pans in the study area, were transformed through the reclamation and utilization of natural wetlands by humans. The area of artificial wetlands showed a trend of substantial increase. The area of breeding ponds increased from 3.34 km 2 in 1986 to 149.47 km 2 in 2017. The area of salt pans has not changed much, and there were occasional staggered distributions in the breeding pond area. By 2017, there were no salt pans, and they were all used by breeding ponds, agriculture, and industry.
In the non-wetland landscape types, farmland, woodland, and construction land continued to increase. Among them, farmland and woodland increased from 0 to 76.55 km 2 in 2017. Since 1986, construction land grew from 0 to 93.5 km 2 in 2017, a significant increase.

Change Rate and Dynamic Degree of the Nantong Coastal Wetland Types
The dynamics of each single landscape type of coastal wetlands in four time periods were statistically analyzed using Formula 1 from the Section 2. The results (Table 4) showed that during the 30 years from 1986 to 2017, the Suaeda salsa salt marsh had the highest dynamic degree, with an average annual change rate of −20%, especially in the period of 2002-2008, where the reduction degree reached 53.9%. The second was nonwetland types, such as construction land. The average annual change rate of construction

Change Rate and Dynamic Degree of the Nantong Coastal Wetland Types
The dynamics of each single landscape type of coastal wetlands in four time periods were statistically analyzed using Formula 1 from the Section 2. The results (Table 4) showed that during the 30 years from 1986 to 2017, the Suaeda salsa salt marsh had the highest dynamic degree, with an average annual change rate of −20%, especially in the period of 2002-2008, where the reduction degree reached 53.9%. The second was non-wetland types, such as construction land. The average annual change rate of construction land was 8.4%. The average annual change rate of farmland, woodland, and other landscapes was 7.9% and exceeded 10% in some periods. Artificial wetlands also had a high degree of dynamics. The average change rates of breeding ponds and salt pans were 4.1% and 6.2%, with some periods exceeding 10%.

Landscape Index Analysis
Number of patches and patch density: As shown in Figure 5, from 1986 to 2017, the number of patches in the bare mudflat and offshore, Suaeda salsa, reed, thatch, and breeding ponds increased first and then decreased. In particular, reed, thatch, and Suaeda salsa decreased significantly in the past decade, from 254 to 51 in total. The number of patches of river banks, Spartina salt marshes, and farmland showed a trend of 'increase-decreaseincrease.' The amount of land for construction and other uses increased significantly, from Mean patch area and largest patch index: As shown in Figure 6, from 1986 to 2017, the average patch area of river banks and salt pans increased first and then decreased; Suaeda salsa, construction land, reed, thatch, and Spartina salt marshes decreased first and then increased; and breeding ponds as well as farmland continued to increase, from 4.69 to 276.41 and from 0 to 107.76, respectively. The largest patch index of construction land, river bank, farmland and woodland, breeding ponds, and others showed a significant upward trend. However, the largest patch index of bare mudflat and offshore as well as Suaeda salsa salt marshes continued to decrease, from 95.33 to 84.98 and from 0.62 to 0.04, respectively, while other landscape types changed their volatility. Fractal dimension index: The larger the fractal dimension of the landscape patch, the more complex the edge of the patch. As shown in Figure 7, except for bare mudflat and offshore, the average fractal dimension of other natural wetlands was higher than 1.06. The reason for this may be that Spartina alterniflora, Suaeda salsa, and reed have similar ecological niches, while Spartina alterniflora has a growth advantage and gradually invades other wetland communities. The average fractal dimension of artificial wetlands was higher. In general, the average fractal dimension of each landscape type over the 30 years was higher than 1.05.  Shannon's diversity index and evenness index: As shown in Figure 8a,b, alth the diversity changes were small from 1986 to 1993, at 0.02, the range of changes incr from 1993 to 2017, from 0.08 to 0.21, indicating that the differences in various we landscape types widened and that the degree of landscape heterogeneity increased change in the landscape diversity index better reflected the process of rapid increa the intensity of human activities such as reclamation and port construction in the Na coastal zone. The change trend in the evenness index was basically the same as that diversity index, showing an increasing trend from 1993 to 2017, from 0.12 to 0.31 distribution of different wetland landscape types became increasingly uniform, an difference in proportional structure gradually decreased.
Contagion index and aggregation index: As shown in Figure 8c,d, from 1986 to the value of the contagion index fluctuated within a small range, from 0.26 to 0.30 landscapes in the study area were relatively fragmented, with many small patches were mainly affected by a few large landscape types (bare mudflat and offshore, bre ponds, etc.). The aggregation index showed a continuous decreasing trend, from 98 97.63, indicating that the aggregation degree between patches of different landscape continued to weaken and that the fragmentation degree increased. This is consistent the findings of the diversity index and evenness index, mainly due to the disturban human activities and construction. Shannon's diversity index and evenness index: As shown in Figure 8a,b, although the diversity changes were small from 1986 to 1993, at 0.02, the range of changes increased from 1993 to 2017, from 0.08 to 0.21, indicating that the differences in various wetland landscape types widened and that the degree of landscape heterogeneity increased. The change in the landscape diversity index better reflected the process of rapid increases in the intensity of human activities such as reclamation and port construction in the Nantong coastal zone. The change trend in the evenness index was basically the same as that of the diversity index, showing an increasing trend from 1993 to 2017, from 0.12 to 0.31. The distribution of different wetland landscape types became increasingly uniform, and the difference in proportional structure gradually decreased.
Contagion index and aggregation index: As shown in Figure 8c,d, from 1986 to 2017, the value of the contagion index fluctuated within a small range, from 0.26 to 0.30. The landscapes in the study area were relatively fragmented, with many small patches, and were mainly affected by a few large landscape types (bare mudflat and offshore, breeding ponds, etc.). The aggregation index showed a continuous decreasing trend, from 98.73 to 97.63, indicating that the aggregation degree between patches of different landscape types continued to weaken and that the fragmentation degree increased. This is consistent with the findings of the diversity index and evenness index, mainly due to the disturbance of human activities and construction.

Influence Factors
The landscape changes in the Nantong coastal wetlands are mainly controlled by three factors: one is the natural geographical conditions, the second is the demand for land, and the third is the difference in economic comparative interests. The latter two are both human interferences. In different periods, these three factors have different impacts on the changes in the Nantong coastal wetland landscape, and, in many cases, they are complementary and inseparable from each other.

Influence of Natural Conditions
As far as natural conditions are concerned, there were two main factors that caused the differences in the structure and pattern of the Nantong coastal wetlands landscape. On the one hand, the difference in the types of coastal erosion and deposition resulted in the difference in the abundance and shortage of reserve resources for coastal land development and utilization. This caused the north-south difference in the landscape structure, especially the landscape structure with natural covering as the main body. Nantong is dominated by long silt coasts, especially the coast north of Dayanggang. The salt marsh vegetation coverage in this area is high, and the main vegetation types grow in patches, especially in the estuary area, such as Donglinggang and Yaowanggang. From the shore to the sea, the patchy distribution of reed-thatch-Suaeda marshes changed from the patchy distribution of reed-thatch-Spartina salt marshes. On the other hand, ecological succession caused changes in the landscape pattern of coastal wetlands. The coastal ecosystem itself has a natural succession process: From the sea to the land, presenting a cross-sectional distribution of 'subtidal zone-intertidal zone (bare mudflat)-Spartina-Suaeda salsa-reed and thatch.' Corresponding to this section, the elevation of the coastal wetlands increased slightly, the soil salinity gradually decreased, and the plant types with different salt tolerances and flood tolerances grew in segments, thus forming different landscape types.

Influence Factors
The landscape changes in the Nantong coastal wetlands are mainly controlled by three factors: one is the natural geographical conditions, the second is the demand for land, and the third is the difference in economic comparative interests. The latter two are both human interferences. In different periods, these three factors have different impacts on the changes in the Nantong coastal wetland landscape, and, in many cases, they are complementary and inseparable from each other.

Influence of Natural Conditions
As far as natural conditions are concerned, there were two main factors that caused the differences in the structure and pattern of the Nantong coastal wetlands landscape. On the one hand, the difference in the types of coastal erosion and deposition resulted in the difference in the abundance and shortage of reserve resources for coastal land development and utilization. This caused the north-south difference in the landscape structure, especially the landscape structure with natural covering as the main body. Nantong is dominated by long silt coasts, especially the coast north of Dayanggang. The salt marsh vegetation coverage in this area is high, and the main vegetation types grow in patches, especially in the estuary area, such as Donglinggang and Yaowanggang. From the shore to the sea, the patchy distribution of reed-thatch-Suaeda marshes changed from the patchy distribution of reed-thatch-Spartina salt marshes. On the other hand, ecological succession caused changes in the landscape pattern of coastal wetlands. The coastal ecosystem itself has a natural succession process: From the sea to the land, presenting a cross-sectional distribution of 'subtidal zone-intertidal zone (bare mudflat)-Spartina-Suaeda salsa-reed and thatch.' Corresponding to this section, the elevation of the coastal wetlands increased slightly, the soil salinity gradually decreased, and the plant types with different salt tolerances and flood tolerances grew in segments, thus forming different landscape types.

Impact of Reclamation
With population growth and economic development, in the late 1960s and 1970s, in order to solve the contradiction between more people and less land, the movement of reclaiming land from the sea with the main purpose of increasing the area of arable land arose. By the 1990s, more than 140 km 2 of tidal flats had been reclaimed in the Nantong coastal zone [50]. However, due to poor development conditions and low economic benefits, the reclamation speed was not fast, and the scale was not large. In the early 1990s, Nantong actively responded to Jiangsu's strategic idea of 'Construction of Maritime Sudong' and, accordingly issued some preferential policies for the development of coastal tidal flats, such as special concessions for developers in areas with poor locations and resource conditions. As a result, the development activities of the Nantong coastal zone have also entered a new stage. On the one hand, in order to develop tidal flats, the construction of infrastructure in coastal areas has been further strengthened. On the other hand, diversified management and the development of tidal flats have been implemented, and the construction of supporting facilities for reclamation development and utilization (mainly irrigation channels, roads, etc.) has been increased. From 1993 to 2002, the area of breeding ponds increased by more than 20 times. Since 2009, a new round of tidal flat reclamation and port development as well as construction has set off a new round of climaxes. The mode of utilization has also changed significantly, from the main form of agriculture and fishery to a combination of agriculture, fishery, and industrial parks. These development patterns around the comprehensive utilization of tidal flats will inevitably affect the landscape structure and pattern of the Nantong coastal wetlands ecosystem.
In addition to reclamation, port development as well as expansion are also some of the main reasons for the loss of coastal wetlands and changes in landscape patterns. The construction of Yangkou Port, Tongzhouwan Port, and these coastal industrial parks have all become significant factors in the changing of the land use of the Nantong coastal areas in this century. The construction of ports isolates the wetlands into small habitat patches, cutting off the connection between them and interfering with the normal migration of organisms.
Finally, the human-induced invasion of alien species has a huge impact. Since 1982, Jiangsu coastal wetlands have developed into the largest Spartina alterniflora salt marsh ecosystem in China. The coastal area of Nantong was the earliest area where Spartina was introduced. The growth and spread of Spartina alterniflora has had a very significant impact on the landscape pattern and changes in its coastal wetlands.

Difference in Economic Comparative Interests
Differences in comparative interests in land use are another important cause of changes in coastal landscape patterns. The characteristic of coastal landscape change is that it evolves in the direction of maximizing land-use value because under the situation of economic diversification, people gradually realize that diversification can bring more lucrative benefits. For example, in the late 1970s, the utilization of the Nantong coastal zone was still dominated by agriculture. By the 1980s, it appeared that grain and cotton production in addition to aquaculture were the mainstays, and the comprehensive development and scale management of agriculture, forestry, animal husbandry, fishery, and salt were implemented. The utilization of tidal flats has gradually developed in a diversified direction, and the development benefits have been greatly improved. Over the years, a production and utilization pattern of the coexistence of ports, grain and cotton, shrimp, eel, freshwater fish, forest fruit, animal husbandry, clams, seaweed, and reeds has been gradually formed to maximize the comparative benefits of land. With the diversification of land use, the fragmentation of the coastal zone landscape increases, and the landscape presents a landscape change trend in multiadvantage development.

The Importance and Future of the Nantong Coast
Silty sand muddy coasts are widely distributed in China, accounting for about onethird of the coastline. Among them, the Nantong coast of Jiangsu is the most typical area. Therefore, the trend and magnitude of landscape change in the Nantong coastal area represent the changing characteristics of coastal wetlands along the silty sand muddy coast of China. The Nantong coastal wetlands are characterized by rich biodiversity, high productivity and ecological effects, and high potential economic value. The unscientific land reclamation and coastal development activities in the past decades have led to a significant decrease in natural wetlands and a significant increase in non-wetland land-use practices. In response to this prominent problem, the National Development and Reform Commission and the Ministry of Natural Resources jointly issued the Master Plan for Major Projects for the Protection and Restoration of Nationally Important Ecosystems (2021-2035), which includes major projects for the ecological protection and restoration of coastal wetlands as a major strategy. The Nantong coast will also carry out specific projects for the protection and restoration of coastal wetlands in the future. This study will provide a scientific basis for specific coastal wetland protection and restoration projects and can provide scientific support for the preparation and implementation of the comprehensive protection and utilization plan of the coastal zone in the national coastal area.

Deficiencies and Prospect of Our Study
In this study, we used remote sensing images (five representative time points) combined with field surveys to classify landscape types, and achieved credible results, but there are still shortcomings to be improved. First, for time-series, we need to add more remote sensing images and obtain time-series results with shorter intervals, such as a 1-year interval, which will be more beneficial for us to observe more details of landscape changes. Second, in terms of remote sensing interpretation methods, the application of high-level classification methods, such as deep learning, have higher accuracy than traditional methods, such as supervised classification. In the next step, we will combine field survey data to build a database of remote sensing interpretation samples for various landscape types to support the development of reliable intelligent interpretation methods and provide a data base for subsequent landscape change analysis. Again, there is little information on the underwater topography in this study area, and the specific location of the −6 m isobath needs to be corroborated by more field measurements. Therefore, the change of underwater topography is not well considered in the area statistics. In the next step, we will expand the remote sensing interpretation method of underwater topography and revise the research scope of coastal wetlands in different periods to reduce the error.

Conclusions
Taking the coastal wetlands in Nantong, Jiangsu, China as the study area, combining field investigation and remote sensing interpretations, we extracted wetland types from the five-period remote sensing images from 1986 to 2017. A single dynamic model and multiple landscape indices method was used to analyze the variation characteristics of various types of coastal wetlands.
The main findings are as follows: (1) The Nantong wetland type system was established, which was divided into three major categories: natural wetland landscapes, artificial wetland landscapes, and nonwetland landscapes, including 11 subcategories (reed marsh, thatched marsh, Suaeda salsa salt marsh, Spartina salt marsh, bare mudflat and offshore, river bank, breeding pond, salt pan, construction land, farmland and woodland, and others (abandoned land, reclamation of unused land, etc.)). (2) Natural wetlands, such as thatched and Suaeda salsa marshes, were extremely reduced, while artificial wetlands and non-wetland with high human activity, such as breeding ponds, farmland, and construction land, increased significantly in the Nantong coast.
The two types of natural wetland vegetation showed opposite trends as a whole in the past 30 years. The area of salt marshes of Suaeda salsa shrunk rapidly, while the area of salt marshes of Spartina increased significantly.
(3) In the past 30 years, due to the influence of environmental pressures such as population growth, land demand, and economic development, the major influencing factors of local landscape change shifted from natural geographical factors to human activities and economic as well as social factors. With the diversification of land use, the fragmentation of the coastal zone landscape increased, and the landscape presented a landscape change trend of multiadvantage development. Therefore, it is worth noting that reduced human activities and increased conservation as well as restoration efforts should be implemented to bring a stable increasing trend to the Nantong coastal wetlands.