Response of Landscape Evolution to Human Disturbances in the Coastal Wetlands in Northern Jiangsu Province, China

: Human disturbance is one of the essential driving forces of landscape evolution. The quantitative evaluation of the spatial and temporal characteristics of landscape evolution and its relationship with human disturbance are of great signiﬁcance to regional ecological protection and management and are crucial for achieving coordinated socioeconomic development and ecological– environmental protection. In this study, we took the coastal wetlands in northern Jiangsu province, China, as the research area, and proposed a quantitative evaluation method for directional landscape evolution. On this basis, the spatiotemporal characteristics of the landscape evolution from 1980 to 2020 and the relationship with human disturbance were quantitatively evaluated by combining a human disturbance index and statistical methods. The results showed that: (1) The area of the natural wetlands decreased signiﬁcantly over the past 40 years, while the areas of artiﬁcial wetlands and non-wetlands increased signiﬁcantly. (2) The landscape evolution process was dominated by the degradation process. The main types of degradation were natural wetland conversion to artiﬁcial wetland and non-wetland areas and Spartina alterniﬂora invasion. The restoration type was mainly restoration among artiﬁcial and natural wetlands. (3) The degradation of wetland landscapes demonstrated a southward shift trend and the spatial consistency with the change of the human disturbance index was high (the correlation coefﬁcient was 0.89). (4) The human disturbance index was signiﬁcantly and positively correlated with the rate of degradation, with a correlation coefﬁcient of 0.43, and was not signiﬁcantly and positively correlated with the restoration rate, with a correlation coefﬁcient of 0.14. The ﬁndings in this paper provide additional information and theoretical guidance for the control of coastal wetland development and utilization, as well as for achieving coordinated wetland resource development together with utilization and ecological protection in the coastal wetlands of Jiangsu province, China.


Introduction
A landscape is a heterogeneous geographical unit consisting of different ecosystem types over a certain extent of land [1]. The landscape pattern determines the form of the spatial distribution of its resources and environment and influences the regional ecological processes and functions. Coastal wetlands are important habitats and breeding grounds for offshore organisms and migration transit stations for birds, and are precious wetland resources with important ecological functions, such as water conservation, water purification, soil and water conservation, climate regulation and biodiversity protection functions [2,3]. They are also intersection areas for four major domains (the lithosphere, atmosphere, hydrosphere, and biosphere), with frequent material-energy exchange, as well as transition zones of sea-land ecosystems, with high degrees of vulnerability and sensitivity [2][3][4]. At the same time, coastal areas display active socioeconomic development,

Landscape Data
Landsat MSS/TM/OLI data for 1980,1985,1990,1995,2000,2005,2010,2015, and 2020 were used in the study. This is mainly because these eight periods correspond well with China's 5-year plan and the development stages of the study area, which is in line with the stages of coastal wetland development, which can highlight the stages of landscape evolution more. The landscape data were interpreted by researcher Guosheng Li and Dr. Linlin Cui from the Institute of Geographical Sciences and Resources, Chinese Academy of Sciences; associate researcher Jilong Chen from the Chongqing Institute of Green Intelligence, Chinese Academy of Sciences; and senior engineer Huajun Liao from SuperMap Software Co., Ltd., using a visual interpretation method based on Landsat data (a total of eight periods of landscape data). Based on the field survey data in 2015 and the 1:200,000 beach resource survey data in 1980s, the accuracy of the interpreted data in 2015 and 1980 was evaluated as being between 81. 45 and 91.67%, respectively. Since the data for other years were applied with the same interpretation method as in 2015, the accuracy of the interpretation data was found to be relatively reliable and met our application requirements. The specific method can be found in the literature [32], and some of the data have been used in the literature [2,18,32]. In addition, the landscape data showed that the mean patch area of the study area over the eight target years was 1.83-3.06 km 2 , so 30 m Landsat pixels were sufficient for interpretation.
The landscapes included three major categories based on the National Wetland Resources Survey and Monitoring Technical Regulations (China): natural wetlands, artificial The landscapes included three major categories based on the National Wetland Resources Survey and Monitoring Technical Regulations (China): natural wetlands, artificial wetlands, and non-wetlands, among which the natural wetlands were divided into shallow marine waters, tidal flats, Suaeda glauca, Phragmites australis, grasslands, rivers, and Spartina alterniflora; the artificial wetlands were divided into pools, salt fields, aquafarms, and paddy fields; the non-wetlands were divided into drylands, woodlands, barelands, urban settlements, rural settlements, levees, and other construction lands (as shown in Figure 2).

Landscape Evolution Evaluation Method
The directional landscape evolution model was proposed in 2015 by Cui et al. [2,18], which combines landscape evolution with directional discrimination for the evaluation of the wetland landscape evolution directionality ( Figure 3). In this study, the landscape evolution direction was divided into the restoration direction and the degradation direction, while the evolution within non-wetlands was not part of this study. wetlands, and non-wetlands, among which the natural wetlands were divided into shallow marine waters, tidal flats, Suaeda glauca, Phragmites australis, grasslands, rivers, and Spartina alterniflora; the artificial wetlands were divided into pools, salt fields, aquafarms, and paddy fields; the non-wetlands were divided into drylands, woodlands, barelands, urban settlements, rural settlements, levees, and other construction lands (as shown in Figure 2).

Landscape Evolution Evaluation Method
The directional landscape evolution model was proposed in 2015 by Cui et al. [2,18], which combines landscape evolution with directional discrimination for the evaluation of the wetland landscape evolution directionality ( Figure 3). In this study, the landscape evolution direction was divided into the restoration direction and the degradation direction, while the evolution within non-wetlands was not part of this study.

Landscape Evolution Evaluation Method
The directional landscape evolution model was proposed in 2015 by Cui et al. [2,18], which combines landscape evolution with directional discrimination for the evaluation of the wetland landscape evolution directionality ( Figure 3). In this study, the landscape evolution direction was divided into the restoration direction and the degradation direction, while the evolution within non-wetlands was not part of this study.  The direction of natural wetland landscape degradation includes the following changes: (1) natural wetlands converted into artificial wetlands or non-wetlands; (2) natural community succession involving soil dewatering; (3) natural community succession involving soil salinity accumulation; (4) the invasion of exotic species. Regarding the artificial wetlands, the direction of degradation moves as follows: paddy fields → aquafarms → salt fields → pools. The direction of wetland landscape recovery or restoration follows the inverse trend.
Based on the directional evolution model of the coastal wetland landscape in Figure 3, the degraded or restored evolved area was counted in combination with the transfer matrix method and the percentage of degraded or restored evolved area to the total area of the region was used as the evaluation index of wetland degradation or restoration. where K denotes the proportion of the area where degradation or restoration occurred in the region; D i denotes the area where degradation or restoration occurred in the i-th landscape; D is the total area of the region. The theoretical basis of the transition matrix is a quantitative calculation of the transformation between different landscapes. Usually the rows in the transfer matrix represent landscapes at the moment of T 1 and the columns represent landscapes at the moment of T 2 ; A ij represents the total area of landscape i converted to landscape j during T 1 − T 2 ; A ii represents the area where landscape i remained unchanged during T 1 − T 2 .

Human Disturbance Index
Previous studies used the grid method to calculate the human disturbance index, while the integrity of the natural geographic units and the attribution of administrative units were not considered in the grid division. To facilitate data management, we used administrative districts as the statistical unit to calculate the human disturbance index, which was calculated as follows [10,11].
where HI is the human disturbance index of an administrative unit; n is the number of landscapes in an administrative unit; S i is the area of type i landscape within an administrative unit; S is the total area of an administrative unit; C i is the human disturbance coefficient corresponding to the type i landscape. The human disturbance coefficient is a quantitative characterization of the degree of disturbance of human activities with the natural property characteristics of the regional landscape. As the forms and objects of human activities in different regions are different, the size of the disturbance coefficient should be determined according to the purposes of human activities and the characteristics of the object (landscape), and a targeted quantitative setting should be made. Under the scenario of no human activity interference at all, the study area was considered a typical salt marsh wetland landscape. Therefore, the definition and setting of the degree of human disturbance in the region in this paper mainly considered the degree of the disturbance level of human activity on the regional wetland landscape and wetland function. Based on the results of previous research [7,8,[33][34][35][36][37], and combined with the directional landscape evolution model, the greater the deviation of the current landscape from the natural wetland landscape (and function; non-wetland deviation was the largest, followed by artificial wetland), the greater the defined disturbance coefficient, and conversely the smaller the defined disturbance coefficient.
According to the above ideas, the deviation of the non-wetland landscape from the natural wetland landscape (and function) was the largest in the current landscape, and the artificial wetland landscape was the second largest. Therefore, for the convenience of the study, the human disturbance coefficient of the study area was set to six levels (from 0 to 5) in this paper (Table 1). Among them, shallow marine waters and Spartina alterniflora communities are the least affected by human activities due to their inherent properties and geographic distribution characteristics; thus, their disturbance coefficient was set to 0. The tidal flats, Suaeda glauca, Phragmites australis, grasslands, and rivers are natural wetlands, which are less affected by human activities, so the human disturbance coefficient was set to 1. Most of the woodlands in the study area were planted forests and most of the barelands were perennially unexploited lands. Although these are non-wetland in nature, the actual degree of human disturbance is greater than that of natural wetlands and less than that of artificial wetlands; therefore, they were set to 2. Most of the pools were artificially excavated, and thus the disturbance coefficient was set to 3. Paddy fields, aquafarms, salt fields, and drylands are the most significant types of landscapes with planting and farming activities in this study area, and the degree of human disturbance is large; therefore, the disturbance coefficient was set to 4. Urban settlements, rural settlements, levees, and other construction lands are the non-wetland landscapes most affected by human development and construction activities, with the strongest interference coefficient, which was set to 5.

Statistical Analysis
To analyze the trends in the area of landscape evolution and the anthropogenic disturbance index, regression analysis was applied in this study. It is an important method to study the trends of variables in time series, based on the time variable x and the landscape area y, which can be described as follows [38]: where a is the slope, b is the intercept, and ε is the random error. The linear equation is obtained by least squares linear fitting and the slope a indicates the trend of landscape area y values. The goodness of fit (R 2 ) test and F-test were used to test the effectiveness of the regression analysis. R 2 is the degree of fit of the regression line to the observations. The closer the value of R 2 is to 1, the better the fit of the regression line to the observed values; conversely, the smaller the value of R 2 is, the worse the fit of the regression line to the observed values. Statistical significance was set at the 0.05 level. The Pearson correlation coefficients between the human disturbance index and the rate of landscape evolution, the landscape evolution area, and the change quantity of the landscape index were calculated in order to analyze the effects of human disturbance on landscape evolution and evaluate the reliability of the landscape evolution analysis method.

Landscape Indices
To evaluate the reliability of the landscape evolution analysis method, on the basis of previous studies [11,39,40], considering the structural characteristics of the coastal wetland landscape in the study area and the ecological significance of the landscape index, the mean patch area (Area-MN) and aggregation index (AI) were used in this study at the landscape level. The specific calculation formula is as follows: (1) Area-MN is commonly used for landscape fragmentation evaluation. Smaller Area-MN values represent a more fragmented landscape: where PA is the total area of patches (km 2 ) and PN is the number of patches. (2) AI is used to characterize the degree of aggregation and non-randomness. Larger AI values represent higher landscape aggregation and better connectivity: where m is the total number of patch types; g ii is the number of nodes between pixels of patch type i based on the single-fold method; p i is the area ratio of the landscape comprised of patch type i.

Analysis of Area Change
In the past 40 years, the decrease in wetlands in the study area was 529.87 km 2 , with a rate of decrease of 13.25 km 2 /year. The natural wetlands showed a significant decreasing trend (R 2 = 0.94, p < 0.05), while artificial wetlands (R 2 = 0.93, p < 0.05) and non-wetlands (R 2 = 0.97, p < 0.05) showed significant increasing trends ( Figure 4). The natural wetlands decreased by 2344.83 km 2 , accounting for 33.18% of the total study area (7068.22 km 2 ), with an average annual decrease rate of 58.62 km 2 /year ( Table 2). The main types of natural wetlands that decreased were tidal flats (1637.08 km 2 ), Suaeda glauca (476.94 km 2 ), grasslands (342.92 km 2 ), and Phragmites australis (136.96 km 2 ), accounting for 63.11%, 18.39%, 13.22%, and 5.28% of the net decrease in area of the above four types of natural wetlands, respectively. The total increase in artificial wetlands was 1814.96 km 2 , accounting for 25.68% of the total study area, with an average annual increase rate of 45.37 km 2 /year. The main increase was 1699.10 km 2 in aquafarms, followed by 402.55 km 2 in paddy fields and 9.88 km 2 in pools, accounting for 93.62%, 22.18%, and 0.54% of the net increase in the area of artificial wetlands, respectively (the area of salt paddies decreased by 296.57 km 2 , accounting for 16.34%).    The area of coastal wetlands in the study area accounted for 23.98% of the 10 coastal administrative districts (Xiangshui county, Binhai county, Sheyang county, Dafeng city, Dongtai city, Hai'an city, Rudong county, Tongzhou district, Haimen city, and Qidong city); however, the changes in the natural and artificial wetlands accounted for 86.96% and 70.30% of the changes in the natural and artificial wetland areas in the 10 coastal administrative districts, respectively. The loss of natural wetland area and the increase in artificial wetland area in the coastal area of Jiangsu province, China, in the past 40 years mainly occurred in the coastal wetland area, while the rate of natural wetland loss was higher than the rate of artificial wetland increase. In addition, Spartina alterniflora was the Remote Sens. 2021, 13, 2030 9 of 21 only natural wetland landscape with an increasing trend in area. This area increased by 238.45 km 2 in the past 40 years, accounting for 3.37% of the total study area of 7068.22 km 2 , with an average annual rate of increase of 5.96 km 2 .

Analysis of the Landscape Evolution Process
The change area of natural wetlands was the largest (3111.76 km 2 ), accounting for 82.44% of the total change area (3774.50 km 2 ), while the change area of artificial wetlands accounted for only 13.99% (Table 3). Among the transformation processes of natural wetlands, the internal conversion among natural wetlands, and the conversion of natural wetlands to artificial wetlands and non-wetlands accounted for 23.00%, 62.76%, and 14.24% of the total conversion area of natural wetlands, respectively. The transformation of natural wetlands in the study area was the most significant change in the past 40 years. In particular, the conversion of natural wetlands to artificial wetlands was the most obvious. During 1980-2020, the total area of landscape transformation was 3774.50 km 2 , accounting for 53.40% of the total area of the study area. The area of landscape degradation reached 3042.03 km 2 , accounting for 43.04% of the total area of coastal wetlands and 44.52% of the area of the wetlands in the previous period (1980). The area of landscape restoration reached 697.92 km 2 , accounting for 9.87% of the total area of coastal wetlands, 10.21% of the area of wetlands in the previous period (1980), and accounting for 22.94% of the degradation ( Table 3). The sum of the degraded and restored evolved area of the landscape accounted for 99.08% of the total evolved area of the landscape. The evolution of modern wetland systems in the study area is still dominated by the degradation process, and the degree of restoration and wetland restoration is relatively low. The areas of natural wetlands → artificial wetlands, natural wetlands → non-wetlands, and Spartina alterniflora invasion equaled 1953.07, 443.07, and 329.05 km 2 , respectively accounting for 64.20%, 14.56%, and 10.82% of the total degraded area. The degradation of wetlands in the study area over the past 40 years-mainly the transition type of natural wetlands to artificial wetlands and non-wetlands, as well as the invasion of Spartina alterniflora-also requires attention. The degradation of artificial wetlands (artificial wetlands → non-wetlands and the degradation within artificial wetlands) covered 223.36 km 2 , accounting for 7.34% of the total degraded area. This indicates that the degradation of artificial wetlands was weak. Among the types of restoration, those among natural and artificial wetlands accounted for the highest proportions, covering 293.09 and 266.61 km 2 , respectively, and accounting for 41.99% and 38.20% of the restoration area in the region, respectively. This indicates that the evolution of wetland landscape restoration in the study area in the past 40 years was mainly based on the gradual restoration of natural wetlands, such as soil desalination, while the conversion of salt fields into aquafarms under the intervention of human activities was also one of the important reasons for the structural restoration of wetland systems in the study area.
The landscape degradation area in the study area from 1980 to 2020 showed a nonsignificant decreasing trend as a whole and had obvious phasing characteristics. There was a non-significant increasing trend in the degradation area from 1980 to 2000 and a non-significant decreasing trend in the degradation area from 2000 to 2020 ( Figure 5). Except for the degradation among natural wetlands, which showed a significant monotonic decline, all other types showed a non-significant increasing trend, followed by a decreasing trend, and the inflection points were all in the 1995-2000 period. accounted for the highest proportions, covering 293.09 and 266.61 km 2 , respectively, and accounting for 41.99% and 38.20% of the restoration area in the region, respectively. This indicates that the evolution of wetland landscape restoration in the study area in the past 40 years was mainly based on the gradual restoration of natural wetlands, such as soil desalination, while the conversion of salt fields into aquafarms under the intervention of human activities was also one of the important reasons for the structural restoration of wetland systems in the study area. The landscape degradation area in the study area from 1980 to 2020 showed a nonsignificant decreasing trend as a whole and had obvious phasing characteristics. There was a non-significant increasing trend in the degradation area from 1980 to 2000 and a non-significant decreasing trend in the degradation area from 2000 to 2020 ( Figure 5). Except for the degradation among natural wetlands, which showed a significant monotonic decline, all other types showed a non-significant increasing trend, followed by a decreasing trend, and the inflection points were all in the 1995-2000 period. The main degradation types differed in the different time periods; however, the proportion of natural wetlands → artificial wetlands, natural wetlands → non-wetlands, and Spartina alterniflora invasion to the total degraded area ranged from 51.53% to 85.16%. This indicates that these three types were the main wetland degradation types in the study area. In response to the landscape restoration, the sudden decrease in the area of restoration in the 2015-2020 time period may be related to the reduction in wetland restoration under human disturbance from strengthening of the wetland protection in recent years.
The 1980-2015 landscape restoration in the study area showed an overall non-significant upward trend and also had segmental characteristics. The inflection point was also located in the 1995-2000 time period ( Figure 6). The main types of restoration were natural and artificial wetland inter-restoration, both accounting for more than 58% of the restoration, except for the 2000-2005 and 2015-2020 time periods. The main degradation types differed in the different time periods; however, the proportion of natural wetlands → artificial wetlands, natural wetlands → non-wetlands, and Spartina alterniflora invasion to the total degraded area ranged from 51.53% to 85.16%. This indicates that these three types were the main wetland degradation types in the study area. In response to the landscape restoration, the sudden decrease in the area of restoration in the 2015-2020 time period may be related to the reduction in wetland restoration under human disturbance from strengthening of the wetland protection in recent years.
The 1980-2015 landscape restoration in the study area showed an overall non-significant upward trend and also had segmental characteristics. The inflection point was also located in the 1995-2000 time period ( Figure 6). The main types of restoration were natural and artificial wetland inter-restoration, both accounting for more than 58% of the restoration, Combined with the results in Table 3, although the evolution of modern wetland systems in the study area from 1980 to 2020 was still dominated by degradation processes and the degree of restoration and wetland restoration was relatively low, the degradation was on a decreasing trend and the restoration was on an increasing trend; thus, the wetland ecosystem in the study area as a whole was on a restoration trend.

Spatial-Temporal Characteristics in Landscape Evolution
From Figure 7, the degradation of the coastal wetland landscape in northern Jiangsu, China, in the past 40 years mainly occurred in Sheyang county, Dafeng city, Dongtai city, and Rudong county, where the degraded area accounted for 84.37% of the total degraded area in the region. The degradation types were mainly natural wetlands > artificial wetlands, natural wetlands > non-wetlands, and Spartina alterniflora invasion. The proportions of degraded areas in each administrative district from largest to smallest were Dafeng city (67.14%) > Sheyang county (54.63%) > Dongtai city (52.30%) > Hai'an city (42.10%) > Qidong city (38.05%) > Rudong county (28.70%) > Haimen city (25.46%) > Binhai county (22.49%) > Tongzhou district (20.59%) > Xiangshui county (19.04%). The landscape restoration in the study area mainly occurred in Sheyang county, Binhai county, Dafeng city, Rudong county, and Ringshui county. The restoration area accounted for 84.98% of the total restoration area in the region, while the restoration types were mainly the restoration among artificial wetlands and the restoration among natural wetlands. The proportions of restored areas in each administrative district were Binhai county (35.21%) > Sheyang county (18.77%) > Xiangshui county (16.05%) > Tongzhou district (10.70%) > Dafeng city (7.86%) > Qidong city (7.78%) > Hai'an city (7.38%) > Rudong county (4.21%) > Dongtai city (3.73%) > Haimen city (3.63%). Combined with the results in Table 3, although the evolution of modern wetland systems in the study area from 1980 to 2020 was still dominated by degradation processes and the degree of restoration and wetland restoration was relatively low, the degradation was on a decreasing trend and the restoration was on an increasing trend; thus, the wetland ecosystem in the study area as a whole was on a restoration trend.
Tables 4 and 5 display the rates of degradation and recovery evolution of the coastal wetland landscape areas for eight periods in each county and city, respectively. The rates of degradation from Xiangshui county to Dafeng city in the north showed a decreasing trend, while those from Dongtai city to Qidong city (except Rudong county) in the south showed an increasing trend, indicating a southward trend of coastal wetland landscape degradation in northern Jiangsu, China. Remote Sens. 2021, 13, x FOR PEER REVIEW 13 of 23 13 Figure 7. The spatial distribution pattern and types of wetland landscape evolution in the past 40 years.
Tables 4 and 5 display the rates of degradation and recovery evolution of the coastal wetland landscape areas for eight periods in each county and city, respectively. The rates of degradation from Xiangshui county to Dafeng city in the north showed a decreasing trend, while those from Dongtai city to Qidong city (except Rudong county) in the south showed an increasing trend, indicating a southward trend of coastal wetland landscape degradation in northern Jiangsu, China. Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with a decreasing trend), decreasing-increasing trend (rate of degradation with a decreasing trend and recovery evolution with an increasing trend), increasing-increasing trend (rate of degradation with an increasing trend and recovery evolution with an increasing trend), and increasing-decreasing trend (rate of degradation with an increasing trend and recovery evolution with a decreasing trend). Three counties and cities in the study area had a decreasing-decreasing trend for the rate of degradation and restoration of coastal wetlands in the last 40 years. The absolute value of the slope of the rate of degradation was larger than that of the rate of restoration and the coasts of the three counties are all siltation coasts, which indicates that the ecological conditions of these counties and cities were slowly improving and that more attention should be paid to them in the future. Two counties and cities showed a decreasing-increasing trend, for which the recovery evolution rates in the latter periods were greater than the rates of degradation, indicating that the ecological condition of these two counties and cities was improving faster. However, since the shorelines of Xiangshui and Binhai counties are of the erosion type, the coastal wetlands of these two counties and cities should also receive continuous attention. There were three counties and cities with increasing-increasing trends, where the shorelines of Haimen and Qidong were relatively stable. The slopes of their rates of degradation and recovery evolution were close and the average rate of degradation was larger than the rate of recovery evolution, which indicates that the ecology of coastal wetlands in Haimen and Qidong will continue to deteriorate. The slope of the rate of degradation in Dongtai city, which is located on a siltation-type coast, was five times that of the slope of the recovery evolution, while the average rate of degradation was two times that of the recovery evolution, which indicates that the coastal wetlands in this city were experiencing accelerated degradation. Table 4. Changes of the coastal wetland areas' rates of degradation in 10 counties and cities in the study area over eight periods.        Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with    Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with Table 5. Changes of the coastal wetland areas' rates of restoration in 10 counties and cities in the study area over eight periods.      Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with  Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with   Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with   Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with     Considering the trends for the rates of degradation and recovery evolution in the eight periods, the 10 counties and cities were classified into four categories: decreasingdecreasing trend (rate of degradation with a decreasing trend and recovery evolution with

Region
In the future, more resources should be invested in these three counties and cities, and effective measures should be taken to control the degradation and to promote the evolution of the restoration. Although the area of coastal wetlands in these two counties is relatively small, policymakers should pay greater attention to them due to the accelerated degradation of the coastal wetland landscape in this region.

Correlation Analysis of the Human Disturbance Index and the Rate of Landscape Evolution
The human disturbance index of the coastal wetlands in the study area showed a significant monotonic increasing trend (p < 0.05) from 0.91 in 1980 to 2.25 in 2020 (Figure 8a). The trends of human disturbance in Yancheng city and Nantong city were the same as those in the study area, while the values and rates of change of human disturbance index in Yancheng city were greater than those in Nantong City.
The changes of the human disturbance index in all 10 counties and cities in the study area showed a significant upward trend and were ranked in order of the mean value of the human disturbance index: Xiangshui county > Binhai county > Sheyang county > Dafeng city > Dongtai city > Hai'an city > Qidong city > Rudong county > Tongzhou district > Haimen city (Figure 8b).
The amount of variation in the human disturbance index in Xiangshui county, Binhai county, Sheyang county, Dafeng city, and Rudong county showed a decreasing trend. The amount of variation in the human disturbance index in Dongtai city, Hai'an city, Tongzhou district, Haimen city, and Qidong city showed an increasing trend (Figure 8c).
Although the human disturbance index in the study area showed a significant increasing trend, due to the differences in the size of the human disturbance base in each county and city, the base in the northern counties and cities was large and the human disturbance gradually stabilized with a decreasing trend, while the human disturbance base in the southern counties and cities was small and showed a faster growth trend. This indicates a southward trend of human disturbance.
In terms of the amount of change, the human disturbance index was significantly and positively correlated with the rate of degradation throughout the study area, with a correlation coefficient of 0.43 (p < 0.05). The human disturbance index was not significantly and positively correlated with the rate of restoration, with a correlation coefficient of 0.14 (p > 0.05) (Figure 8d). There was a strong correlation between wetland landscape degradation and human activities in the study area, while the correlation between restoration and human activities was weak.
For each county and city, except for Rudong county, Haimen city, and Qidong city, the human disturbance and the rate of wetland landscape degradation in the rest of the counties and cities were positively correlated. In particular, Sheyang county and Dafeng city were highly significantly correlated. The human disturbance index and the rate of wetland restoration in each county and city were insignificantly correlated; Sheyang county and Dafeng city were positively correlated, while the rest of the counties and cities were negatively correlated.
From the spatial pattern of change, the human disturbance had a strong spatial consistency with the degradation and restoration-the slope of change of the human disturbance factor was significantly and positively correlated with the rate of degradation and the rate of restoration for all counties and cities, with correlation coefficients of 0.89 (p < 0.05) and 0.68 (p < 0.05), respectively.

Human Distribution, Natural Change, and Landscape Evolution
Disturbance is an important ecological process that includes both natural and human disturbances. Human disturbance mainly affects the landscape structure, function, and diversity by changing landscapes (e.g., evolving from low-disturbance natural or semi-natural landscapes to high-disturbance anthropogenic landscapes) and the spatial distribution and configuration of their components [11,[41][42][43]. The history of landscape evolution in the coastal wetlands in northern Jiangsu, China, comprises coastal wetland resource development driven by natural geographical features (natural conditions) plus socioeconomic development, with prominent stage and spatial variability characteristics. The coastal wetland is a good land reserve resource and Jiangsu province has abundant coastal wetland resources with a long history of polder reclamation, which can be traced back to the construction of the Fan Gong dike [33]. Since 1980, the development and utilization of coastal wetlands has changed from sea salt development, demolishing salt-making stoves and cultivation land, to land reclamation and cotton cultivation to the comprehensive development and utilization of land for agriculture, forestry, animal husbandry, fisheries, and salt [44,45]. The study area has gone through several stages of development-from the development strategy of "Marine Sudong" in the Ninth Five-Year Plan to establish a new grain and cotton base; to the development of millions of mu (1 mu = 1/15 hectare) of mudflats in the 10th Five-Year Plan; to the development of "ports, towns, and coastal harbor industry" in the 11th Five-Year Plan; to the construction of a "new port industrial zone, modern agricultural base, new energy base, and coastal new city" in the 12th Five-Year Plan; and finally to the integrated development of a "port, industry, and town" in the 13th Five-Year Plan. This series of economic development initiatives has led to dramatic changes in land use patterns, and the intensity of human disturbance has gradually increased, eventually leading to the evolution of regional landscapes-a large number of natural landscapes (such as tidal flats, Suaeda glauca, and Phragmites australis) have been developed and utilized as aquafarms, salt fields, arable land (paddy fields and drylands), settlements, etc.
The preliminary results of this paper are discussed below. First, there was a southward trend of landscape degradation in the coastal wetlands in the study area (Table 4), while the changes in the human disturbance index also showed a decreasing trend to the north of Dongtai City and an increasing trend to the south (Figure 8). The reasons for the southward landscape degradation trend of the coastal wetlands may be related to the variability of regional physical geographic characteristics (natural conditions) and socioeconomic development. Xiangshui county, Binhai county, and the northern part of Sheyang county (the mouth of the Guanhe River to the mouth of the Sheyang River) are located near the mouth of the abandoned Yellow River and belong to the erosive coast, where the reclamation potential is small. Therefore, the internal evolution of the anthropogenic landscape occurred mainly in this region. From the southern part of Sheyang county to Haimen city (from the mouth of Sheyang River to Dongzao Port), the old Yellow River's submerged deltaic sediment and Yangtze River's inlet sediment (under the joint action of the north-south tidal wave system and the off-shore radiating sandbar) form a siltation environment with less wind and waves. The siltation is strong, which is of typical siltation along the coast. The Qidong city section is part of the stable coastal section [46].
Second, the development of human disturbance intensity demonstrated an obvious stage development process, closely related to the economic development stage in the region; thereby, the coastal wetland landscape degradation produced significant stage characteristics. For 1980-1990, the intensity of the human disturbance in the coastal wetland region was low. Coupled with the emphasis on the polder approach and not utilization, the regional economic development level was low, which made the rate of landscape degradation generally decrease in all counties and cities (Table 4). After 1990, with the population growth and economic development, the urbanization process for coastal Jiangsu accelerated, while the development of coastal wetlands was an important initiative to reduce the economic differences between the north and the south, enhancing the intensity of the reclamation development of coastal wetlands [45]. The rate of landscape degradation from Xiangshui county to Hai'an city reached a maximum in 1995-2000 (Table 4). After 2015, the government strengthened the protection of coastal wetlands by strictly controlling reclamation, while Jiangsu province also introduced a national ecological protection red line plan to further protect coastal wetlands. This led to a rapid decrease in the rate of landscape evolution in the study area from 2015 to 2020.
The above preliminary research results indicate that the northern part of the Jiangsu coast needs to further optimize the landscape, while the southern part needs to strengthen the protection of coastal wetlands to achieve coordination of the coastal wetland development and utilization with ecological protection as far as possible.
Third, the rate of landscape degradation in the study area was significantly positively correlated with the changes in the human disturbance index. This, on the one hand, also supports the judgment that human disturbance may be the driving force of landscape evolution in the study area. On the other hand, it also indicates that landscape evolution may be influenced by certain other factors such as climate change. In estuarine coastal areas, changes in water exchange at the land-sea interface; changes in groundwater levels; and sensitive habitat factors such as soil salinity caused by beach exploitation, global climate change, and sea level changes control the development and evolutionary processes of coastal wetland ecosystems [47]. Global climate change affects the runoff from the estuaries, coastal hydrodynamic conditions, and regional evapotranspiration. The increase in salinity of tidal flats under conditions of reduced inlet runoff inevitably leads to changes in coastal ecological conditions, which affects the growth and development processes of various plants and animals in the soil and water bodies. Under conditions of relatively low precipitation, changing the coastal surface water or groundwater by a few centimeters may degrade the wetland ecology [48]. Sea level increases not only reduce tidal flats and coastal wetlands due to erosion or inundation [49], but can also affect the process of landscape evolution of coastal wetlands by altering the composition and salinity of sediments and the hydrological pattern of coastal wetlands [50,51]. At the same time, sea level increases may also increase the frequency and intensity of natural disasters, such as storm surges, salt water intrusion, and flooding, which indirectly interfere with the ecological processes such as nutrient cycling and primary production of coastal wetland systems, undermine the stability and balance of coastal wetland systems, lead to increased ecosystem vulnerability, and ultimately lead to the degradation of the landscape ecology of coastal wetland systems.
In addition, landscape evolution and human disturbance intensity are scale-dependent, and this paper was directly based on the statistics of 30 m resolution data without considering the influence of different scales on the landscape evolution and human disturbance intensity.

Reliability of the Landscape Evolution Analysis Method
In this paper, we used the directional landscape evolution model to study the landscape evolution of coastal wetlands. To verify the reliability of the method, based on the results of previous studies, we selected Area-MN and AI at the landscape scale to perform a correlation analysis with the landscape evolution area. The results showed that the amount of Area-MN change in Yancheng city, Nantong city, and the whole study area showed a weak negative correlation with the corresponding landscape degradation and restoration areas. The amount of AI change showed a negative correlation with the landscape evolution area in Yancheng city and the study area, and a weak positive correlation with the landscape evolution area in Nantong city. In particular, AI showed a significant negative correlation with the area of landscape degradation in Yancheng city and the study area (Table 6). Table 6. Correlation coefficients between the landscape evolution area and change quantity of the landscape index. Mean patch area (Area-MN) and landscape aggregation index (AI).

Yancheng City
Nantong City Study Area

Restoration Area
Degradation Area

Restoration Area
Degradation Area
From 1980 to 2020, many natural landscapes changed from continuous large patches of natural landscapes with low human disturbance to a series of relatively small patches of anthropogenic landscapes with high human disturbance. The results of previous studies have shown that wetland degradation increases landscape fragmentation and decreases the connectivity of the same landscapes [52][53][54]. The results of the correlation analysis showed that landscape degradation was negatively correlated with Area-MN and AI, which indicates that landscape degradation increased the landscape fragmentation and decreased the landscape connectivity. The results of this paper also corroborate the findings of previous studies, which to a certain extent also proves the reliability of the method proposed in this paper. In addition, the landscape evolution demonstrated a non-significant weak negative correlation with Area-MN, especially in Nantong, which showed a nonsignificant weak positive correlation with AI, which also indicates that the landscape index needs to be chosen carefully.
In addition, it should be noted that the landscape evolution model, which is the core of the directional landscape evolution evaluation method proposed in this paper, has no way to be directly applied in other regions due to the regional variability of landscapes in different regions. However, the framework of this method and the construction method of the evolution model are universal.

Uncertainty of the Human Disturbance Index
Zhou et al. (2018) analyzed the spatial and temporal changes of human disturbances in the Jiangsu coastal zone, China, based on 1.5 km × 1.5 km grids, and the results showed that the human disturbance index had an upward trend from 1995 to 2013 [11]. The increments of the human disturbance index were larger in Sheyang County, Dafeng City and Dongtai City in 1995-2002, while the regions with larger increments in 2002-2013 were Qidong City and Haimen City. This indicates a southward trend of the human disturbance. These findings are consistent with the results of the current study. The administrative unit was chosen as the evaluation unit for the human disturbance index in this study. Compared with the standard grid unit, the areas of administrative units presumably vary, which can lead to larger index values for administrative units, with smaller areas when the area of a given landscape is the same. The human disturbance within administrative units is not uniform, and the closer the area to a settlement, the more serious the human disturbance is. This study uses the average disturbance level within administrative units instead of disturbance within counties and cities, which may be different from the actual situation. The human disturbance index based a standard-sized grid unit may be more reasonable for studying its spatial distribution characteristics. At the same time, we are not clear about the influence of different spatial scales on the relationship between human disturbance and landscape evolution. Therefore, the scale effects of human disturbance and landscape evolution should be further explored to reduce uncertainty through the integration of multi-scale information, with a view to better supporting coastal wetland restoration, conservation, and management.
The disturbance coefficient is a key parameter for calculating the human disturbance index, and its assignment is based on landscape data, which are highly subjective. The magnitude of the disturbance coefficient only characterizes the relative degree of deviation of different landscapes in the region from the natural wetland landscape and has only relative significance. Therefore, the same landscape in different areas may represent different intensities of human activities or a certain landscape in the same area with different patch characteristics (area, size, density and shape, etc) may also represent different intensities of human activities. Therefore, further research is still needed on how to assess the human disturbance coefficient more scientifically and objectively in a quantitative manner.

Conclusions
In this paper, we quantitatively analyzed the spatial and temporal characteristics of landscape evolution and its response to human disturbance in the coastal wetlands in northern Jiangsu, China, from 1980 to 2020 through a directional landscape evolution model and localized human disturbance coefficients.
The results of the study showed that from 1980 to 2020, landscape degradation was the main feature of regional landscape evolution. In 1980-1990, due to the emphasis on the polder approach and not utilization, the economic efficiency was not high, which resulted in a reduced rate of landscape degradation for this period. From 1990 to 2005, the rate of landscape degradation reached its maximum value due to the accelerated urbanization process and the increased economic benefits of reclamation development and utilization. After this, the rate of landscape degradation gradually decreased as the difficulty and cost of polder reclamation increased and the awareness of conservation increased. After 2015, the protection of coastal wetlands was strengthened at the national level and polder reclamation was strictly controlled, while Jiangsu province also introduced the "national ecological protection red line plan", which resulted in the rate of landscape degradation in the study area from 2015 to 2020 decreasing rapidly. At the same time, due to the obvious southward trend of landscape degradation and human disturbance, the southern part of the study area needs to further increase its control over coastal wetland development and utilization, so that the development and utilization of coastal wetlands can be carried out in a coordinated manner with ecological protection.
In this paper, we applied a method to quantitatively evaluate the evolution of the coastal wetland landscape, and through a comparative analysis with the traditional landscape index, we found that this method had a certain degree of reliability. However, we cannot ignore the shortcomings of this study, which are mainly reflected by two aspects. First, the directional landscape evolution model only considered landscape degradation and restoration, ignoring the transformation of land types (transformation between nonwetland types) beyond the directional discrimination of the method. Second, landscape evolution and human disturbance intensity are scale-dependent, and this paper was directly based on the statistics of 30 m resolution data without considering the influence of different scales on the landscape evolution and human disturbance intensity. Third, inconsistencies in the interpretation accuracy of the landscape data may also add to the uncertainty of the results. The above shortcomings should be further explored in future studies to reduce the uncertainty of research conclusions and to better support the promotion of the coordinated development of coastal wetland exploitation and ecological protection.  Data Availability Statement: All satellite data used in the study are available for free download from their respective data portals (https://earthexplorer.usgs.gov/, accessed on 9 April 2021).