Landscape Ecological Risk Assessment of Zhoushan Island Based on LULC Change

Abstract

Owing to limited land resources and unique ecosystems, islands face more serious ecological risks under the interference of climate change and human activities. In this study, selecting Zhoushan Island as the study area, a landscape ecological risk index model was constructed based on LULC (land use/land cover) data and the landscape ecological risks for Zhoushan Island from 2000 to 2020 were analyzed. The results showed that: (1) From 2000 to 2020, the proportion of forest land and grassland remained above 70%, built-up land expanded from 52.67 km² to 123.52 km², and the beach area and ocean on the north side of the island decreased by 23.24 km² and 24.87 km², respectively; this was mainly converted into built-up land. (2) The number of landscape patches in Zhoushan Island decreased as the landscape ecological risk index decreased. The landscape ecological risk showed distinct spatial autocorrelation, with lowest-risk and medium-risk areas collectively accounting for 80% and higher-risk and highest-risk areas showing a decline. (3) The landscape ecological risks exhibited distinct spatiotemporal differences. Before 2010, the higher-risk and highest-risk areas were mainly distributed in the mudflat and ocean areas on the northern coast. After 2010, the higher-risk and highest-risk areas are mainly distributed in the central region, which comprises woodland, grassland, and built-up land.

1. Introduction

Global climate change and irrational human activities have aggravated sea level rises and ocean warming, which profoundly affect coastal zones and island ecosystems [1,2]. As land areas surrounded by the sea and far from the mainland, offshore islands are not only an important part of the marine ecosystem [3] but are also a frontier for developing the marine economy and safeguarding national maritime rights and interests [4]. As islands are affected by both land and sea, their ecosystems are more fragile than terrestrial ecosystems and face greater ecological risks [5].

Ecological risk represents the risk to the population, ecosystems, and even parts of the landscape that constitute an area [6,7]. Landscape pattern is the result of different natural phenomena and human social activities acting at different scales. Accordingly, in larger-scale studies, landscape patterns can reflect the LULC status of a region. We also use RS and GIS to analyze the LULC of the area and to further evaluate the condition of parts of the landscape [8,9]. Landscape ecological risk assessment refers to the quantitative analysis and evaluation of the evolution process and spatial distribution of regional ecological risk from the perspective of landscape pattern [10], and its indicators can accurately express the different states of landscape types within the region. Based on LULC, the landscape ecological risk index has a complete framework and system and has been widely applied by many scholars to large-scale areas such as peninsulas [11], coastal zones [12], and watersheds [13]. It can provide a decision-making basis for scientific risk prevention, the rational use of resources, and strict implementation of environmental protection measures. However, the same human activity has a greater impact on smaller islands. Therefore, changes in the landscape ecological risk indices of islands also need to be explored.

At present, research on islands as geospatial bodies has been conducted focusing on the ecological services of islands [14,15], shoreline changes [16], island management [17], etc.; however, landscape risk has rarely been considered. China is a maritime power country with numerous islands. According to the “China Natural Resources Statistics Bulletin 2022” [18], China has more than 11,000 islands. Therefore, monitoring and studying the changes in LULC and ecological risks on offshore islands is conducive to grasping the development and protection of offshore islands in China, promoting the construction of a marine ecological civilization, and is of great significance to the economical and intensive utilization of land on islands, as well as coordinated land and maritime development.

Zhoushan Island is located on the south side of Hangzhou Bay and is the fourth largest island in China. Since the beginning of the new century, human activities have intensified, the development of Zhoushan Island has accelerated [19], and its LULC has changed frequently, which has also continuously affected its landscape ecological risks. In recent years, with the implementation of marine strategies and the proposal of a marine ecological civilization [20], exploring the spatiotemporal changes of LULC in Zhoushan Island and accordingly conducting a landscape ecological risk assessment have gained importance. This study aimed to gain deeper insight into the above issues through the following research objectives: (1) analyze the temporal and spatial characteristics of LULC change in Zhoushan Island from 2000 to 2020 and (2) quantitatively analyze the temporal and spatial changes of landscape ecological risk in Zhoushan Island using methods such as the landscape ecological risk evaluation index. The basic framework of the study is shown in Figure 1.

2. Data and Methods

2.1. Study Area

Belonging to Zhejiang Province, Zhoushan Island (121°30′–123°25′ E, 29°32′–31°04′ N) is located on the south side of Hangzhou Bay and the west side of the Pacific Ocean; it is the most important island of the Zhoushan Islands. The topography of Zhoushan Island is dominated by low hills with a southeast to northwest trend. Zhoushan Island has a humid subtropical monsoon climate, with an average annual temperature of approximately 16 °C. It also features short and radial rivers flowing into the sea that are surrounded by narrow alluvial plains. From 1995, Zhoushan entered the period of island industrialization [21]. After entering the new century, the development of Zhoushan Island accelerated and the cross-sea bridge was opened to traffic in 2009. Zhejiang Zhoushan Archipelago New Area was officially established in 2011 and Zhoushan Free Trade Port Area was established in 2017, which further promoted the development of Zhoushan. As a typical island city, its urbanization rate has reached 73%, which already corresponds to a high level of urbanization [22]. Therefore, it is of great significance to further balance the LULC and ecological security of the island for the future high-quality development of Zhoushan Island. Considering these issues, Zhoushan Island was selected as a typical and representative research area, and its results can guide the future development and construction of other islands.

2.2. Data Sources

The data used in this study were obtained from the GEE (Google Earth Engine) platform. Considering the growth of surface plants and the interference of image cloud cover, the images from 1 May to 30 September in 2000, 2005, 2010, and 2015 and from 1 February to 31 October 2020 were selected. Among the selected remote sensing images, the ones from the years 2000, 2005, and 2010 were derived from the Landsat 5 TM series, whereas the images corresponding to 2015 and 2020 were obtained from the Landsat 8 OLI series. The spatial resolution of the images used in this study was 30 m.

Based on this, the selected images were de-clouded and synthesized by a mean value algorithm in code on the GEE platform to ensure the best quality of the acquired images [23]. In order to improve the accuracy of sample selection, NDBI (Normalized Difference Build-Up Index), MNDWI (the Modified Normalized Difference Water Index), and NDVI (Normalized Difference Vegetation Index) were constructed as feature variables for the random classification based on different combinations of bands in Landsat imagery, which is a common means of classification accuracy improvement. By referring to the previous literature [24] and combining it with the actual situation of Zhoushan Island from the perspective of land and sea coordination, when combining GEE with Google Earth Pro, a manual visual interpretation method was used and the LULC types during the study period were divided into built-up land, water bodies, woodland, grassland (including arable land), mudflats, and sea areas. The total number of samples for each year was 322, 296, 234, 304, and 118, of which 70% was used for classifier training and 30% for result validation. The total classification accuracies for the five years were 0.90, 0.87, 0.87, 0.87, and 0.88 and the Kappa coefficients were 0.87, 0.84, 0.83, 0.83, and 0.89, respectively. In addition, we calculated the F1 scores obtained from the LULC classification model in this study, which were 0.80, 0.80, 0.82, 0.87 and 0.93 for the five individual years. Combined with the overall classification accuracy, kappa coefficient, and F1 scores, this indicates that the current data meets the needs of this study [25].

2.3. Methods

2.3.1. LULC Dynamics

The LULC dynamics can reflect the rate of change in the number of LULC types in the study area. In this study, a single LULC dynamic and a comprehensive LULC dynamic degree were used to measure the degree of LULC transformation in Zhoushan Island during different periods [26].

The formula for calculating the single LULC dynamics is:

$$K=\frac{U_b-U_a}{U_a}\times \frac{1}{T}\times 100\%$$

where K is the index of a certain type of LULC dynamics during the study period (%) and U_a and U_b are the areas at the beginning and end of a study period for an LULC type, respectively, with T being the number of years studied.

The formula for calculating the comprehensive LULC dynamics is:

$$LC=\frac{{\displaystyle {\sum }_{i=1}^n\Delta L{U}_{i-j}}}{2{\displaystyle {\sum }_{i=1}^{n}L{U}_i}}\times \frac{1}{T}\times 100\%$$

where LU_i is the area of the LULC type I at the start of detection, and in this case: LU_i is the area of the LULC type I at the start of the monitoring time (km²); ∆LU_i−j is the absolute value (km²) of the area converted from type I to non-type I during the monitoring period; and T is the monitoring period (years). When the time period T is set to a year, the value of LC is the annual rate of change for an LULC type in the study area (%).

2.3.2. Analysis of LULC Transfer Matrices

The transfer matrix method is derived from the quantitative description of system state and state transition in system analysis. Using the land transfer matrix form, the transfer changes between various types of land types can be clearly understood [24]. The mathematical form of this matrix is:

$${\mathrm{S}}_{\mathrm{i}\mathrm{j}}=\left|\begin{array}{cccc}{\mathrm{S}}_{11}& {\mathrm{S}}_{12}& \dots & {\mathrm{S}}_{1\mathrm{n}}\\ {\mathrm{S}}_{12}& {\mathrm{S}}_{22}& \dots & {\mathrm{S}}_{2\mathrm{n}}\\ \dots & \dots & \dots & \dots \\ {\mathrm{S}}_{\mathrm{n}1}& {\mathrm{S}}_{\mathrm{n}2}& \dots & {\mathrm{S}}_{\mathrm{n}\mathrm{n}}\end{array}\right|$$

where S represents area, n represents the number of types of LULC, and i and j represent the LULC types at the beginning and end of the study period, respectively.

2.3.3. Division of Landscape Ecological Risk Assessment Units

The division of risk assessment units is an important step in landscape ecological risk assessment. In order to calculate the landscape ecological risk index, the evaluation unit of Zhoushan Island was divided using the fishnet tool in ArcGIS.

Based on the area and computational complexity of the study area and the research results of previous scholars in the field of landscape ecology [27], the research area was divided into 0.5 × 0.5 km evaluation units, and a total of 2195 evaluation units were obtained (see Figure 2). Using Fragstats software (Version 4.2), the landscape ecological risk index value of each unit was calculated and then it was assigned as an attribute to the central point of each evaluation unit [8].

2.3.4. Construction of Landscape Risk Index

The landscape index is not only the condensation of landscape pattern information but also the key to constructing the landscape ecological risk index and carrying out the quantitative analysis of ecological landscape risk [28]. In this study, the landscape loss index of Zhoushan Island was calculated from the landscape disturbance index and landscape vulnerability index, and the landscape ecological risk index was constructed with reference to previous studies [12,28]. The formula is as follows:

$$ERI={\sum }_{i=1}^n\frac{A_{ki}}{A_k}\times R_i$$

where R_i is the landscape loss index, which represents the area of the landscape type i in the kth risk community, Ak represents the total area of the kth risk community, and n is the landscape type. ERI_i is the landscape ecological risk index of risk community i. The larger the ERI value, the higher the degree of ecological risk and vice versa. The specific indices for calculating the ERI are shown in

Table 1. Calculation formula and meaning of related indicators.

Landscape Index	Expression and Formula	Meaning
Landscape Loss Index (Ri)	$R_i=E_i\times V_i$, Ei and Vi represent the landscape disturbance index and landscape vulnerability index, respectively.	Ri reflects the interaction of external influences on the landscape with its own characteristic [12].
Landscape Disturbance Index (Ei)	$E_i=a{C}_i+b{N}_i+c{F}_i$ a, b, c are the weights of each index, Ci, Ni, and Fi represents the landscape fragmentation index, landscape separation index, and landscape fractal dimension.	Ei is a quantitative expression of the degree of disturbance of different landscapes. The higher the value, the greater the degree of disturbance. In this study, the weight coefficients a, b, and c are 0.5, 0.3, and 0.2, respectively [27].
Landscape Fragmentation Index (Ci)	$C_i=\frac{n_i}{A_i}$ ni represents the number of patches of landscape; Ai represents the total area of landscape of type i.	Landscape fragmentation is a process in which the internal properties of the landscape gradually become complex for various reasons, forming various heterogeneous and discontinuous patch mosaics [8]. This index indicates the degree of fragmentation of the landscape; the larger the value, the higher the degree of fragmentation.
Landscape Separation Index (Ni)	$N_i=\frac{A}{{2A}_i}\sqrt{\frac{n_i}{A}}$ A represents the total area of the landscape.	Ni represents the degree of dispersion in the spatial distribution of different patches of a landscape type [29]. The higher this value, the more chaotic the separation.
Landscape Fractal Dimension (Fi)	$F_i=\frac{2\times \ln \frac{l_i}{4}}{\ln A_i}$ Li represents the perimeter of landscape type i.	Fi means the fractal dimension index of the internal geometry of the patch, with larger values indicating more complex landscape patch structures and variations [30].
Landscape Vulnerability Index Vi	Landscape vulnerability is an intrinsic factor affecting regional ecological risks and refers to the resilience of landscapes to external risk activities [8]. In this paper, six different landscape types are assigned and the assigned values are normalized according to the assigned value: from smallest to largest, built-up land 1, woodland 2, grassland 3, waterbodies 4, sea areas 5, mudflats 6. The normalized vulnerability coefficients are respectively 0.0476, 0.0952, 0.1429, 0.1905, 0.2381, and 0.2857.

.

2.3.5. Spatial Autocorrelation Analysis

Spatial autocorrelation analysis can be used to determine whether a feature has a spatial aggregation relationship by describing whether a feature’s attribute values are related to spatially adjacent features [31]. In this study, ArcGIS was used to calculate the global Moran’s index, which was used to determine the occurrence of a spatial aggregation relationship in the landscape ecological risk of Zhoushan Island. Moreover, the local Moran’s index was calculated to measure the spatial aggregation mode of the landscape ecological risk index of Zhoushan Island.

The formula for calculating the global Moran’s index is:

$$I_{G\mathrm{lobal}-\mathrm{Moran}’\mathrm{s}}=\frac{{\displaystyle \sum _i^n{\displaystyle \sum _j^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}}}{{{\displaystyle s}}^2{\displaystyle \sum _i^n{\displaystyle \sum _j^n{w}_{ij}}}}$$

The formula for calculating the local Moran’s index is:

$$I_{\mathrm{Llobal}-\mathrm{Moran}’\mathrm{s}}=\frac{\left(x_i-\overline{x}\right)}{{{\displaystyle s}}^2}{\displaystyle \sum _{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)}$$

where n is the number of research units, x_i and x_j are the observations of spatial positions i and j, respectively, S² is the variance of the score values, and w_ij is the spatial weight matrix.

3. Results and Analysis

3.1. Spatial–Temporal Analysis of LULC Change in Zhoushan Island

3.1.1. Spatial Distribution Characteristics of LULC and Changes in Use Structure

The LULC structure and its changes can reflect human activities and economic development in a region [32]. Figure 3 shows the LULC distribution of Zhoushan Island from 2000 to 2020. In the five time nodes (2000, 2005, 2010, 2015, and 2020) of the study, the percentages of the sum of woodland and grassland areas were 77.37%, 78.88%, 72.89%, 81.99%, and 73.30%, all of which exceeded 70% and thus occupied a dominant position in the LULC type of Zhoushan Island. In 2000, the sea area was mainly distributed on the north and east sides of Zhoushan Island, covering an area of 26.49 km². By 2020, the sea area had decreased by 24.87 km², with only 1.62 km² remaining in the northeast of Zhoushan Island. In 2000, the mudflat was mainly distributed in a strip along the northern coast of Zhoushan Island, with an area of 24.30 km². However, in 2020, it decreased by 23.24 km² to only 1.06 km².

In 2000, the built-up land of Zhoushan Island was 52.67 km², and by 2020, the built-up land had expanded to 123.52 km², an increase of 2.35 times. The expansion of built-up land mainly occurred in the flat terrain and two reclamation areas around Zhoushan Island (Diaoliang Reclamation Area and Donggang Reclamation Area). During the study period, the water area of Zhoushan Island did not change much, from 11.93 km² in 2000 to 9.89 km² in 2020, a decrease of only 2.03 km², which is more stable than the other LULC types. Controlled by the topography of the island, which has a high middle and low surroundings, there are no large rivers inside our study area; mountain ponds and reservoirs primarily account for the water body area. As water is a necessary requirement for people’s daily life production activities, water bodies are likely to be protected from human activities.

3.1.2. LULC Dynamics Analysis and Transfer Analysis

Land-use dynamics can reflect the magnitude of change in a particular land-use type in a given region [26]. In order to gain a more comprehensive understanding of the degree of LULC change in Zhoushan Island, the single LULC dynamics and comprehensive LULC dynamics of Zhoushan Island during the study period were calculated according to the previous formula, as shown in

Table 2. Attitudes towards LULC dynamics on Zhoushan Island (unit: %).

		2000–2005	2005–2010	2010–2015	2015–2020
Single LULC dynamics	Built-up land	−5.26	33.14	−4.80	11.52
	Waterbodies	34.80	−13.89	−10.05	19.82
	Woodland	2.94	0.91	−5.73	10.11
	Grassland	−2.54	−5.18	20.01	−11.42
	Mudflat	−11.80	5.63	−19.00	13.33
	Sea area	−0.20	−10.63	−7.28	−15.84
Comprehensive land use dynamics		2.03	3.06	4.66	5.56

.

Table 2 shows that from the perspective of single LULC dynamics, the LULC dynamics of Zhoushan Island varied significantly from 2000 to 2020. The dynamics of single LULC types varied in the four periods, reflecting the frequent degree of LULC changes within the island. Taking 2010 as the cut-off point, from the perspective of comprehensive LULC dynamics, the comprehensive LULC dynamics during 2000–2005 and 2005–2010 were 2.03% and 3.06% and those during 2010–2015 and 2015–2020 were 4.66% and 5.56%, respectively.

Specifically, the characteristics are as follows: (1) The dynamics of each single LULC type in 2005–2010 and 2015–2020 were not significantly higher than those in other periods, but the comprehensive dynamics were very high, indicating that the interconversion of local types in these two periods was more severe than that in the other periods. (2) The dynamics of built-up land in the 2005–2010 and 2015–2020 periods were significantly higher than those of other types of land, indicating the considerable speed of urban expansion in these two periods. (3) The dynamic degree of the sea area was always less than 0, indicating that the sea area remained in a shrinking state, and the period of rapid shrinkage coincided with the expansion of built-up land. Although the amount of change of forest land was large, its area base was large, and the dynamics showed significant changes during each period. (4) From 2000 to 2020, although the dynamics of waterbodies fluctuated widely, their area base was small and the change in area was thus not prominent.

From the previous formula and the areas of LULC types in Zhoushan Island, the LULC transfer of Zhoushan Island from 2000 to 2020 was mapped and Figure 4 was obtained. As shown in Figure 4a, land transfer from 2000 to 2020 primarily occurred in grassland, sea areas, and mudflats, and the transfer-out areas were 118.13 km², 25.27 km², and 23.88 km², respectively, accounting for 56.48%, 12.08%, and 11.42% of the total transfer-out area. The transferred land is mainly built-up land, with an area of 86.84 km² that accounting for 41.52% of the transferred area, followed by woodland (78.72 km²) and grassland (36.52 km²), which account for 37.64% and 17.46%, respectively. For islands, land resources have always been an important constraint on the progress of development. During the study period, a total of 25.27 km² of the sea area of Zhoushan Island was transferred, of which 66.67% was converted into built-up land. From 2000 to 2005, the mudflat area decreased by 14.34 km², accounting for 59.02% of the mudflat area in 2000; it was mainly converted into built-up land and waterbodies. The mudflat on the north side of Zhoushan Island began to be converted into cultivated grassland and built-up land, with Cen Gang and the north side of Baiquan as the two endpoints, passing through Xiaosha, Ma’ao, and Ganlan, showing a trend of belt transfer as a whole. During 2000–2005, the socio-economic level of Zhoushan was low and people began to encroach on the mudflat. Through engineering constructions, the mudflat was transformed into built-up land in order to support production activities. During 2005–2010, Zhoushan Island was in the growth stage of the island city. On the south side of Zhoushan Island, large areas of grassland in the coastal areas of Yancang, Dinghai Old Town, Lincheng, and Shenjiamen were converted into urban built-up land. In addition, people began to “turn to the sea” through reclamation projects. A large area of the sea near Donggang on the east side of the island was converted into built-up land, transforming the coastline into a flat state. A total of 8.41 km² of the island edge sea area was converted into built-up land during this period. At the same time, human activity began to advance into the grasslands on the sub-edge of the island. During this period, 47.18 km² of cultivated grassland was converted into built-up land, accounting for 63.79% of the area transferred to built-up land. In the expansion of built-up land, the main expansion occurred through the expansion of the old city of Dinghai and the reclamation of land in Donggang. From 2010 to 2015, the development of island cities gradually matured. With the demand for building a sea-garden city, the built-up land of Zhoushan Island did not increase but decreased; moreover, restoring vegetation was practiced on some built-up land with a low-utilization rate. Moreover, reclamation and vegetation restoration occurred in some abandoned areas. During this period, the area of forest land decreased by 72.39 km², accounting for 28.63% of the total area of forest land in 2010; it was mainly converted into grassland. At the same time, the marine area was further reduced and the marine area of fishing by beam trawling within the new area of the Zhoushan Islands was further converted into industrial built-up land to prepare for future industrial construction and production. From 2015 to 2020, the island’s built-up land expanded again. Limited by the topography, the expansion of the built-up land of the island penetrated deep from the edge of the island to the interior of the island. During this period, 60.96 km² of grassland was converted to built-up land. At this time, the conversion of grassland into built-up land occurred across a wide range. Except for the core mountain forest area in the middle of the island, almost all the marginal and sub-marginal areas of the island were converted into built-up land. In 2017, the China (Zhejiang) Pilot Free Trade Zone was established in Zhoushan City. Under this, the northern part of Zhoushan Island was planned as a key development area for oil and other bulk commodity trading bases, promoting the re-expansion of land for island construction. With engineering construction activities, the 2.61 km² sea area in the northeast encountered engineering interception and its marine water gradually decreased and transformed into waterbodies.

Based on the LULC structure change, dynamic degree change, and transfer matrix of Zhoushan Island, the following characteristics could be found. (1) Sea area and mudflats: Driven by the requirements of regional development and construction and limited by the lack of flat area, mudflats and part of the ocean were gradually transformed into urban built-up land, promoting the development of island cities. (2) Woodland and waterbodies: The woodland and waterbodies of Zhoushan Island experienced little change, and some woodland and lake reservoirs in the core area of mountain forests continued to remain in their original states. Woodland plays an important role in soil and water conservation and maintenance of the ecological balance of Zhoushan Island. Zhejiang Province attaches great importance to ecological projects such as afforestation and strictly limits the scale of development and construction for ecological land such as forest land in island areas. Moreover, Zhoushan belongs to an island area with insufficient fresh water sources and reservoirs are required to provide fresh water for production and life. (3) Built-up land and grassland: The overall expansion of built-up land exhibited a pattern of deepening towards the interior of Zhoushan Island. Moreover, the grassland of Zhoushan Island decreased and gradually evolved into built-up land and other land types.

3.2. Spatial–Temporal Analysis of Landscape Ecological Risk Changes in Zhoushan Island

3.2.1. Changes in Landscape Indices of Zhoushan Island

As shown in

Table 3. Sub-index of landscape ecological risk index.

Landscape Type	Year	PA/ha	NP	Ci	Ni	Fi	Ei	Vi	Ri
Built-up land	2000	5425.72	6535	0.6536	0.0627	1.2983	0.2787	0.0476	0.0133
	2005	3932.35	3406	0.4678	0.0721	1.3516	0.2921		0.0139
	2010	10,320.60	8863	0.6227	0.0460	1.2952	0.2730		0.0130
	2015	8023.46	7535	0.6187	0.0533	1.3365	0.2835		0.0135
	2020	12,260.66	5589	0.3905	0.0354	1.2354	0.2578		0.0123
Waterbodies	2000	1242.19	2897	0.7081	0.1215	1.4730	0.3314	0.1905	0.0631
	2005	3552.25	11,390	1.0621	0.0712	1.3371	0.2891		0.0551
	2010	1021.48	1868	0.5509	0.1236	1.5568	0.3487		0.0664
	2015	506.96	534	0.1829	0.1093	1.5467	0.3424		0.0652
	2020	1004.62	1051	0.3303	0.1118	1.4894	0.3317		0.0632
Woodland	2000	21,026.35	5294	0.2452	0.0209	1.1186	0.2301	0.0952	0.0219
	2005	24,046.58	4473	0.1936	0.0168	1.1113	0.2273		0.0217
	2010	25,096.26	6335	0.2162	0.0157	1.1466	0.2341		0.0223
	2015	18,027.71	4717	0.2411	0.0237	1.2026	0.2477		0.0236
	2020	27,101.56	4986	0.2025	0.0163	1.1228	0.2295		0.0219
Grassland	2000	18,185.92	6989	0.2848	0.0196	1.1218	0.2303	0.1429	0.0329
	2005	15,833.94	6356	0.3294	0.0263	1.1626	0.2405		0.0344
	2010	12,028.29	11,830	0.4753	0.0268	1.2379	0.2557		0.0365
	2015	23,565.07	7933	0.2340	0.0125	1.1625	0.2363		0.0338
	2020	10,337.70	8937	0.4405	0.0313	1.2381	0.2571		0.0367
Mudflat	2000	2434.01	2016	0.3939	0.0842	1.3760	0.3007	0.2857	0.0859
	2005	1000.13	934	0.1779	0.0861	1.3992	0.3059		0.0874
	2010	1279.06	1448	0.2857	0.0847	1.4567	0.3170		0.0906
	2015	73.67	436	0.1803	0.1795	1.7226	0.3988		0.1139
	2020	115.75	642	0.2992	0.1739	1.6275	0.3781		0.1080
Sea area	2000	2670.3162	1261	0.4822	0.1431	1.4743	0.3382	0.2381	0.0805
	2005	2619.2525	763	0.1252	0.0552	1.2743	0.2716		0.0647
	2010	1238.7937	894	0.3034	0.1427	1.5646	0.3561		0.0848
	2015	787.6216	627	0.2045	0.1088	1.5471	0.3424		0.0815
	2020	164.2063	241	0.0781	0.0970	1.5133	0.3320		0.0791

, the patch area of built-up land increased but the number of patches decreased, indicating that the area of individual patches of built-up land increased from 2000 to 2020. On the whole, the patches of built-up land showed an aggregation phenomenon. The number of water patches decreased by 1846 from 2000 to 2020, and the C_i index of water body also decreased significantly. However, the area of water bodies decreased by only 237.57 ha, indicating that the aggregation status of water bodies gradually tended to be unified from scattered and fragmented dispersion during the study period. Thus, the R_i index of water bodies did not change significantly. The number of patches of woodland varied significantly, decreasing by 308 overall, and the C_i index and N_i index decreased by 0.0427 and 0.0046, respectively, indicating that the landscape also gradually aggregated and the degree of dispersion decreased. Compared with other landscape types, the F_i index remained at a small level, indicating that the landscape shape complexity of woodland was lower and its landscape type was relatively single. Compared with 2000, the number of patches of grassland increased significantly but the area decreased. The C_i index increased significantly and the state of grassland gradually dispersed, showing a clear fragmented distribution. Compared with other landscapes, the N_i index of mudflat and ocean areas was higher, and the number of patches and patch area of both decreased significantly during the study period. Among them, the multidimensional number of mudflats significantly increased and the distribution gradually became more complex during the study period. The F_i index of the ocean did not change much, increasing only by 0.0390.

3.2.2. Temporal Evolution Characteristics of Landscape Ecological Risks in Zhoushan Island

In areas under intensive human activities, the impacts from land surface modification can be reflected in landscape ecological risks. On this basis, after calculating the ecological risk index for each ecological risk community, the kriging interpolation method was used to obtain the landscape ecological risk distribution in the study area. The landscape ecological risk index in 2000, 2005, 2010, 2015, and 2020 was divided into five levels by referring to the natural breakpoint method provided by the previous literature [28]. ArcGIS: lowest-risk areas (ERI ≤ 0.012), lower-risk areas (0.0112 < ERI ≤ 0.024), medium-risk areas (0.02 < ERI ≤ 0.036), higher-risk areas (0.036 < ERI ≤ 0.048), and highest-risk areas (0.048 ≤ ERI). The results are presented in Figure 5 and

Table 4. Proportion of areas at risk at all levels from 2000 to 2020.

Ecological Risk Level	Area (km2)					Proportion (%)
	2000	2005	2010	2015	2020	2000	2005	2010	2015	2020
Lowest risk	8.02	3.65	8.17	7.89	22.97	1.57	0.72	1.60	1.55	4.51
Lower risk	231.73	235.11	322.94	213.73	395.60	45.47	46.12	63.35	41.93	77.60
Medium risk	225.34	234.69	157.41	283.96	90.44	44.22	46.04	30.88	55.70	17.74
Higher risk	27.18	25.15	9.58	3.95	0.70	5.33	4.93	1.88	0.78	0.14
Highest risk	17.33	11.14	11.65	0.23	0.05	3.40	2.19	2.29	0.05	0.01

.

As shown in Table 4, from 2000 to 2020 the lowest- and lower-risk areas of Zhoushan Island increased by 2.94% and 32.13%, respectively, and the medium-risk, higher-risk, and highest-risk areas decreased by 26.48%, 5.19%, and 3.39%, respectively. From Figure 5, the following results can be observed: (1) During the study period, the area with the highest risk was the smallest, not exceeding 5%. From 2000 to 2020, the area of highest risk decreased by 17.28 km², accounting for 3.39% of the total study area. (2) The area of higher risk (Table 4) has gradually decreased over the past 20 years. From 2000 to 2020, the area decreased from 5.33% to 0.14%, a total decrease of 5.19%. (3) The area of medium-risk initially decreased, then increased, and then decreased, from 44.22% in 2000 to 30.88% in 2010. It later increased to 55.70% in 2015 and decreased to 17.74% in 2020. (4) Between 2000 and 2020 the lower-risk area and the lowest-risk area increased from 45.47% and 1.57% to 77.60% and 4.51%, respectively. (5) In general, among the regional landscape ecological risk level areas in the study area in the past 20 years, the proportions of areas with the highest-risk and higher-risk were small and there was a trend of decline. The lower-risk and medium-risk areas accounted for a relatively large proportion and the area of lowest-risk accounted for a relatively small proportion; however, they showed an upward trend.

3.2.3. Spatial Evolution of Landscape Ecological Risks in Zhoushan Island

The results show that the spatial distribution of landscape ecological risks in Zhoushan Island reflects the distribution law of the landscape structure to a certain extent, with regional and heterogeneous characteristics. As shown in Figure 5, in 2000 the distribution of lowest-risk areas was smaller, mainly in the old city area of Dinghai on the southwest side of the island. Dinghai Old Town has a long history of development, with stable landscape types within the region, and it has formed a stable landscape structure over the years, resulting in low landscape vulnerability and low landscape sensitivity.

In 2000, 2005, and 2010, the highest-risk and higher-risk areas were distributed on the north coast. At this time, the landscape types along the north shore were mainly large areas of mudflats and oceans. The ecological substrate of the two landscapes, mudflats and oceans, is fragile and strongly affected by human activities. Moreover, they have high landscape fragility. By 2015, the mudflat and ocean areas were transformed into land, with more manual intervention, reduced consistency within the landscape, a reduced degree of fragmentation of the landscape, and a unification of landscape types, prompting the decline of the landscape ecological risk index on the north coast and in the northeast. Limited by the topography of Zhoushan Island, the central part is covered by a large area of woodland interspersed with grassland. The woodland and grassland landscape types are susceptible to human factors and pose medium risk. However, controlled by the underdeveloped traditional planting industry in Zhoushan Island and the good ecological environment of Shanghai Island, the medium- and lowest-risk areas continued to be dominant from 2000 to 2015.

After 2015, the further promotion of island urbanization increased human demand for land, human activities began to advance to the sub-edge of the island and the interior of the island and the outer part of the central mountain forest gradually transformed into built-up land. Under the coercion of other risk levels, areas of medium risk and lower risk gradually occupied the entire island and the interior of the island began to show a patchy distribution of ecological risk. The landscape structure of the central mountain forest area was relatively perfect but was also the main area of deforestation and land expansion, with certain potential risks.

By 2020, lowest-risk areas reappeared in the east and south of the island, and the lowest-risk areas remained in areas with perfect urban construction. Urban construction areas themselves were relatively complete, with high stability and low heterogeneity and fragmentation of landscape types. Accordingly, the ecological risk level of the landscape was low.

According to Figure 6, from 2000 to 2020 the landscape ecological risk index decreased in a total of 218.06 km² and increased in a total of 24.06 km². During the study period, the ecological risk level of Zhoushan Island changed mainly in the central core mountain forest area. According to Figure 6, the landscape ecological risk index in the edge area of Zhoushan Island decreased and the increase of landscape ecological risk index showed a phenomenon of patching.

Combined with the actual situation of Zhoushan Island, it can be seen that during the research period the LULC mode in the edge area of Zhoushan Island gradually shifted to unified planning. LULC became more efficient and the degree of landscape fragmentation decreased. At the same time, some areas were converted into built-up land, thereby reducing the fragility of the landscape, increasing the ability to resist external interference, and thus reducing the ecological risk level of the landscape.

3.3. Spatial Correlation Analysis of Landscape Ecological Risks

According to the spatial distribution of landscape ecological risks in Zhoushan Island from 2000 to 2020, Moran’s index of landscape ecological risks in the study area was calculated using ArcGIS, and the results are shown in

Table 5. Global Moran’s Index correlation values.

Year	Moran’s I Index	p Value	Variance	Z Score
2000	0.446468	<0.0001	0.000121	40.678154
2005	0.432329	<0.0001	0.000121	39 391520
2010	0.473517	<0.0001	0.000121	43.164383
2015	0.403730	<0.0001	0.000121	36.784233
2020	0.428432	<0.0001	0.000121	39.029934

.

The table shows that the global Moran’s I index during the study period was greater than 0 and the p value was less than 0.01. The results indicate that the random aggregation pattern of the landscape ecological risk index of Zhoushan Island was less than 1%. This shows that the landscape ecological risk index of Zhoushan Island is positively correlated with the geographical environment in space. In other words, the risk value of areas around an area with high landscape ecological risk of the evaluation unit is also high and the risk value of areas around an area with low landscape ecological risk is also low. Therefore, it can be inferred that the distribution of landscape ecological risks in Zhoushan Island is relatively concentrated and the degree of fragmentation and separation is low. Table 5 shows that this aggregation was most significant in 2010 and lowest in 2015. From the perspective of ecological construction, it is necessary to improve the control of the junction of landscape ecological risks to prevent the spread of risks between regions. In this manner, the increase in landscape ecological risks can be controlled.

As the global Moran’s I index cannot show the spatial correlation and agglomeration characteristics of landscape ecological risk, ArcGIS was used to analyze the landscape ecological risk of Zhoushan Island and the LISA cluster map of landscape ecological risk index was generated (Figure 7). From 2000 to 2020, Zhoushan Island primarily exhibited the high–high aggregation mode, followed by the low–low aggregation mode, with the lowest correspondence to the high–low mode. In 2000, 2005, and 2010, the high–high aggregation pattern of Zhoushan Island was mainly distributed in the northwest mudflat and northeast ocean areas. These two landscapes have a weak substrate and are extremely vulnerable to human activities. In 2015 and 2020, the highest-risk area of Zhoushan Island shifted to the middle of the island and the peripheral area evolved into a lowest-risk area. This phase was the result of an intensification of development activity within Zhoushan Island and renewed urban expansion in the region. The low–low agglomeration areas of Zhoushan Island were mainly distributed on the edge of the island. Combined with the topography of Zhoushan Island and the characteristics of the island, production and construction activities first appeared at the edge of the island, which has a long history of development, mature landscape types, and serious landscape homogeneity in the unit evaluation community. Moreover, its artificial surface is not susceptible to disturbance by external forces and the landscape type is stable. Therefore, a low and low aggregation mode easily forms.

4. Discussion

4.1. Response Relationship between LULC and Landscape Ecological Risks

The relationship between regional ecological risks has always been a research hotspot in the academic community and LULC can directly reflect landscape conditions [32]. Combined with LULC change and the landscape ecological risk of Zhoushan Island, the landscape ecological risk index of Zhoushan Island appears to be declining; however, this does not contradict the traditional impression of “ecological deterioration caused by development and construction”. As the ecological risk of landscape is presented according to the type and structure of the landscape, it is a problem to be faced in the future of the regional landscape. At the beginning of the study period, large areas of the ocean and mudflats were directly exposed, and these two landscape types have fragile substrates and are extremely vulnerable to external disturbances, accompanied by high vulnerability and high risk indices of high exposure. Similarly, taking built-up land as an example, the biodiversity of built-up land is not as high as that of mountains and forest land. However, the stable form of built-up land rarely changes and the degree of ecological risks faced by residents is not large. In contrast, forest land has higher biodiversity. Therefore, in the face of the same external interference, forest land is subject to higher risks than built-up land, and its landscape fragility is thus stronger.

During the study period, Zhoushan Island experienced many major events that promoted regional development and a series of policy implementations promoted the construction of docks, ports, and reclamation areas that changed the landscape type at the edge of the island and reduced the ability of ecological communication between the island and the ocean. Regional development interacts with land-use change, which in turn promotes development. In the mountains far from the edge of the island, people began to expand the scope of activities and construction, alleviating the contradiction between regional development and insufficient land resources. Therefore, in the process of urbanization, appropriate land resources should be reserved to allocate space for the future development of the city.

Within the five study nodes, a significant shift was observed between different risk levels in 2010. The landscape ecological risk index of the study area was higher during 2000–2010 but decreased in 2015 and 2020. In general, human activities had a significant impact on LULC patterns and landscape patterns in Zhoushan Island throughout the study period. The changes and transfers between landscape types and boundaries were more complex and the changes in landscape ecological risks tended to be more complex.

4.2. Comparison with Existing Studies

The landscape ecological risk index is an important index for measuring regional ecological risk [33], and some scholars have conducted relevant research on coastal cities in Zhejiang Province [28] and the coastal zone of eastern China [12]. Wen Zhang [12] pointed out that the ecological risk index of China’s eastern coastal cities decreased significantly from 2000 to 2015. Jialin Li [28] also pointed out that between 1990 and 2010, the main high-risk growth areas of coastal cities in Zhejiang were mainly located on muddy coasts. In this study, the mudflat and ocean areas in Zhoushan Island were identified as the highest-risk areas at the beginning of the study period. With the shrinkage of these landscapes, the landscape vulnerability and risk level also decreased. This is probably consistent with the conclusions of Wen Zhang and Jialin Li.

Jingwen Ai [27] divided Haitan Island into 500 × 500 m assessment units, noting that the overall threat level of landscape ecological risk decreases when large areas of farmland and forests are replaced by impervious land. Areas with the highest level of landscape ecological risk on Haitan Island also declined from 2000 to 2020, which coincides with the results of this study. The decline of the landscape ecological risk index of Zhoushan Island from 2000 to 2020 reflects the stability of the landscape pattern and the gradual maturity of the development status of island-type cities. In this study, the landscape ecological risk assessment of Zhoushan Island was mainly based on LULC. Furthermore, the assessment results are close to the actual situation on Zhoushan Island. However, as Zhoushan is an island city, the scope of this study is the actual extent of Zhoushan Island in 2020. Some areas of similar range, formerly marine areas, have more complex drivers of ecological risk and their evolutionary mechanisms are more difficult to measure.

4.3. Recommendations for Future Development

Zhoushan City is a typical island city in China. In 2000, the Zhejiang provincial government clearly proposed that Zhoushan City should form a belt city with the three groups of Dinghai, Lincheng, and Putuo as the axis and the development direction of Zhoushan Island was gradually clarified. According to the results of the LULC classification in this study, the strip city pattern of Zhoushan Island essentially took shape in 2010 and the direction for future urban development was laid.

At present, the woodland in the central part of Zhoushan Island remains the core ecological region, and it is also the area that will be most disturbed by human activities in the future. Therefore, the boundary between ecological zones and built-up land should be demarcated to ensure the sustainable development of forest land resources and give full play to the ecological effect of forest land. As part of the characteristic industries of Zhoushan Island, heavy industries such as the petrochemical and shipbuilding industries cause to water, soil, and air pollution. Therefore, it is necessary to strictly monitor the status of sewage discharge and waste discharge in the process of development, optimize the industrial structure, and strive to adopt a “low consumption, high output” development model. In the process of the development and construction of Zhoushan Island, governments at all levels should also seek a balance between development and construction and ecological protection, not only to promote social and economic development but also to protect the ecological environment of the region.

5. Conclusions

Based on the LULC data for 5-year intervals from 2000 to 2020, this study analyzed the landscape ecological risk of Zhoushan Island. To this end, risk assessment units of 0.5 × 0.5 km were adopted as the research scale and Fragstats was used. The following conclusions can be drawn: (1) During the study period, the built-up land of Zhoushan Island doubled, increasing to 123.52 km², the beach and ocean area shrank significantly, the water body area remained stable, and the sum of the forest land and grassland areas remained above 70%, serving as the main type of LULC in Zhoushan Island. (2) The change in the LULC type on Zhoushan Island led to the decrease in landscape patches, the gradual decrease of the C_i index and the N_i index, and the overall decline of the landscape ecological risk index. (3) In general, Zhoushan Island has areas of lower-risk and medium-risk, which collectively account for up to 80% of the total area. By 2020, higher-risk areas and highest-risk areas decreased to 0.14% and 0.01%, respectively, decreasing by 5.20% and 3.39% compared with 2000; ecological risks did not show any upward trend. (4) Landscape ecological risks in Zhoushan Island showed significant spatiotemporal differences. Areas of lower-risk were basically distributed in the coastal marginal areas. Areas of higher-risk and highest-risk were mainly distributed in the mudflat and ocean areas on the north coast of the island until 2010, exhibiting high vulnerability. After 2010, the central part of Zhoushan Island became higher-risk and highest-risk areas.