Remote Sensing Based Conservation Effectiveness Evaluation of Mangrove Reserves in China

Abstract

In recent decades, the mangrove area in China has changed dramatically, and governments have established multiple mangrove protected areas at various levels. However, we know little about the effectiveness of conservation on mangroves on a national scale. In this study, we constructed an evaluation index system for landscape health and proposed a landscape health composite index (LHCI) to characterize the landscape health status of mangroves. Based on the distribution dataset of mangrove forests mangrove in the recent 40 years, we evaluated the conservation effectiveness of mangrove reserves in China from a perspective of landscape health. The dynamics of mangrove areas show that the mangrove area in 83% of the reserves increased after the establishment of reserves. Additionally, the increase in mangrove area in provincial-level, municipal-level, and county-level reserves was higher than that in national-level reserves, and the most significant increase in mangrove area was in Guangxi, followed by Fujian and Hong Kong. The evaluation results show that mangrove reserves have achieved outstanding conservation effectiveness in China, with 43% of the reserves significantly improving the landscape health status of mangroves and 35% of the reserves maintaining good condition. The reserves in Guangxi, Guangdong, and Fujian Provinces showed more significant protective effects. Specifically, the most effective reserves protecting mangroves were the Qi’ao Island reserve, Maowei Gulf reserve, and Enping reserve. This study may provide references for formulating a rapid evaluation method of conservation effectiveness based on remote sensing and promote the scientific management of protected areas and the ecological restoration of mangroves in China.

1. Introduction

Mangroves are essential wetland ecosystems with the characteristics of both terrestrial and marine ecosystems, widely distributed in the intertidal areas of tropical and subtropical regions of the world [1,2]. They have significant social, economic, and ecological values and provide a wide range of ecosystem services, such as water purification, shoreline stabilization, reducing coastal erosion, and maintaining biodiversity [3,4,5]. However, due to unreasonable economic development and human overexploitation in coastal areas, the area of mangroves in China has been drastically decreased since the 1950s [6]. The remaining mangroves are also under pressure from climate change and human activities, such as sea-level rise, biological invasion, seawall construction, and aquaculture [7,8,9]. Irreplaceable in ecological and socio-economic services, mangroves have become important targets for wetland conservation and biodiversity protection in China.

Establishing nature reserves is an effective measure to protect and manage mangrove resources [10]. At present, China has nearly 40 nature reserves with mangroves as the primary protection objects established by the state or local governments. Approximately 64% of mangroves are distributed within the nature reserves, and the coverage of the reserves is still expanding [11]. However, mangrove reserves in China started late and lacked systematic planning and design and management experience. Reserves mainly adopted a “rescuing protection” compulsory conservation policy, and the evaluation of its effectiveness was not given sufficient attention [12]. There are four different and complementary aspects in protected areas effectiveness evaluation, i.e., coverage [13,14], degree of detailed monitoring [15,16], management effectiveness [17,18], and conservation effectiveness [19,20]. Specifically, the coverage is an evaluation based on the scope of protection, the degree of detailed monitoring is an evaluation based on the monitoring method, the management effectiveness is an evaluation based on the management behavior, and the conservation effectiveness is an evaluation based on the protection objects of the protected areas. Previous studies on the effectiveness of reserves paid more attention to the evaluation of management effectiveness. According to preliminary statistics, there are about fifty evaluation methods for management effectiveness, which are mainly conducted through literature research, questionnaire surveys, interviews with managers, or expert scoring [17,21]. However, few studies were conducted to evaluate the conservation effectiveness of reserves, especially for mangrove reserves.

Remote sensing has been widely used in the multi-scale and long time-series monitoring of the ecological environment and natural resources, especially for inaccessible mangrove ecosystems [22,23,24]. Due to the characteristics of timely observations and large-area coverage, remote sensing also provides new perspectives and methods for the evaluation of conservation effectiveness [25,26]. A common method is to analyze the changes in the area indicators related to protected objects based on the land cover data obtained from remote sensing monitoring. For example, Wang et al. [27] tested the conservation effectiveness of 20 reserves in Hainan Island by comparing the changes in forest area in locations within the reserves, adjacent to the reserves, and far outside of the reserves. Jia et al. [28] analyzed the dynamics of mangrove areas in seven national reserves before and after the establishment of the reserves. The results showed that China’s mangrove forest conservation effectiveness is better than other Asian countries, and the mangrove area increased immediately after the reserves were established. Porter Bolland et al. [29] compared the annual deforestation rates of community-managed forests and protected forests and found that community-managed forests presented lower annual deforestation rates and higher conservation effectiveness than protected forests. Other studies also evaluated the conservation effectiveness from the perspective of vegetation productivity based on vegetation index or net primary productivity (NPP) data retrieved from remote sensing. For example, Tang et al. [30] used the normalized difference vegetation index (NDVI) to measure variations in plant productivity in the core, boundary, and surroundings of 1015 reserves over a 25-year period. The results suggested that reserves achieved good conservation effectiveness in protecting vegetation productivity. Zhang et al. [31] evaluated the effectiveness of 11 reserves in protecting the ecological environment by analyzing the differences of the NPP before and after the establishment of the reserves and inside and outside reserves.

Besides the area and vegetation productivity indicators, some researchers also evaluated the conservation effectiveness from a landscape perspective based on landscape metrics. Landscape metrics can objectively represent the spatial structure, configuration, and function of the landscape from different levels and perspectives in a quantitative manner, which has been proven to be a reliable indicator for evaluating the conservation effectiveness of reserves [32,33]. For example, Jia et al. [34] analyzed the dynamics of six landscape indices in the Futian reserve and the Mai Po reserve. The results showed that although the integrity and connectivity of mangrove patches improved in both reserves, the integrity of mangrove patches in Mai Po reserve was always higher than that in the Futian reserve, which indicated higher conservation effectiveness of Mai Po reserve. Lu et al. [35] used eighteen landscape metrics to construct an evaluation index system and evaluated the conservation effectiveness of five national wetland nature reserves in the Songnen Plain based on an information entropy model. They found that the natural wetland area of each reserve declined continuously during the study period, and the conservation effectiveness also showed a decreasing trend. In general, the effectiveness evaluation based on landscape metrics is easier to analyze, compare, and monitor than other qualitative methods. The diversity of landscape metrics (e.g., number, area, shape, distance, and connectivity) also makes the evaluation results more comprehensive and credible [36]. Although the methods mentioned above have been widely used and recognized, previous studies on the conservation effectiveness of mangrove reserves in China mainly focused on reserves at local scales, which were insufficient to understand the protection status of mangrove resources in China [37]. Therefore, it is of great practical significance and scientific value to evaluate the conservation effectiveness of mangrove reserves at a national scale, which can contribute to promoting the restoration of mangrove resources and strengthening the management of the reserves.

In this study, we constructed an evaluation index system for landscape health and proposed a landscape health composite index (LHCI) to characterize the landscape health status of mangroves. Based on China’s mangrove distribution dataset with high classification accuracy from 1978 to 2018, we analyzed the dynamics of mangrove landscape health status in 24 representative reserves and evaluated the conservation effectiveness from the perspective of landscape health. To the best of our knowledge, this is the first national-scale evaluation of the conservation effectiveness of mangrove reserves. The results can provide references for the scientific management and adequate protection of mangroves in China.

2. Materials and Methods

2.1. Study Area

The study area covers six coastal provinces (Hainan, Guangxi, Guangdong, Fujian, Zhejiang, and Taiwan) and one special administrative region (Hong Kong). The geographical range of the study area is 18–29°N and 108–122°E (Figure 1). Yueqing County, Zhejiang Province, and Sanya City, Hainan Province, are the northernmost and southernmost boundaries of mangrove distribution in China, respectively. According to the spatial distribution characteristic of mangroves in China, 24 reserves (including 8 national-level reserves, 6 provincial-level reserves, and 10 municipal- and county-level reserves) with mangroves as the primary protection object were selected from the newest China Nature Reserves List and China Marine Reserve List (

Table 1. The detailed information of the selected reserves in this study.

Reserve No.	Name of Reserve	Abbreviation	Level	Province	Establishment Time
1	Yueqing Ximen Island Marine Special Reserve	Ximen Island	National	Zhejiang	2005
2	Quanzhou Bay Estuary Wetland Nature Reserve	Quanzhou Bay	Provincial	Fujian	2002
3	Jiulong River Estuary Mangrove Nature Reserve	Jiulong Estuary	Provincial	Fujian	1988
4	Zhangjiang River Estuary Mangrove Nature Reserve	Zhangjiang Estuary	National	Fujian	1992
5	Danshui River Estuary Mangrove Nature Reserve	Danshui Estuary	Provincial	Taiwan	1986
6	Mai Po Marshes Nature Reserve	Mai Po	National	Hong Kong	1975
7	Neilingding-Futian Nature Reserve	Futian	National	Guangdong	1984
8	Qiao-Dangan Island Nature Reserve	Qi’ao Island	Provincial	Guangdong	1989
9	Enping Mangrove Nature Reserve	Enping	County	Guangdong	2005
	Taishan Zhenhai Bay Mangrove Nature Reserve				2000
10	Pinggang Mangrove Wetland Nature Reserve	Pinggang	County	Guangdong	2005
	Chengcun Haoguang Mangrove Nature Reserve				2000
11	Dianbai Mangrove Nature Reserve	Dianbai	Municipal	Guangdong	1999
12	Zhanjiang (Tongming Bay) Mangrove Nature Reserve	Tongming Bay	National	Guangdong	1990
13	Zhanjiang (Xinliao–Hean) Mangrove Nature Reserve	Xinliao–Hean
14	Zhanjiang (Gaoqiao–Anpu) Mangrove Nature Reserve	Gaoqiao–Anpu
15	Shankou Mangrove Nature Reserve	Shankou	National	Guangxi	1990
16	Maowei Gulf Mangrove Nature Reserve	Maowei Gulf	Provincial	Guangxi	2005
17	Beilun River Estuary Nature Reserve	Beilun Estuary	National	Guangxi	1990
18	Dongzhaigang Nature Reserve	Dongzhaigang	National	Hainan	1980
19	Huachang Bay Mangrove Nature Reserve	Huachang Bay	County	Hainan	1995
20	Caiqiao Mangrove Nature Reserve	Caiqiao	County	Hainan	1986
	Danzhou Dongchang Mangrove Nature Reserve				1986
21	Xinying Bay Mangrove Nature Reserve	Xinying Bay	County	Hainan	1992
22	Sanya River Mangrove Nature Reserve	Sanya River	Municipal	Hainan	1992
23	Qinglan Mangrove Nature Reserve	Qinglan	Provincial	Hainan	1981

) [38,39].

Due to the particular situations of the establishment and management of reserves in China, most reserves have not announced clear boundaries. In this study, based on the location of mangrove reserves and the spatial distribution characteristics of mangroves, we comprehensively delineated the boundary of the key mangrove protection areas in each reserve for evaluating the conservation effectiveness of mangrove reserves in China. In addition, because the spatial distribution of mangroves in the Zhanjiang Reserve, Guangdong Province, was too scattered, we divided it into three zones with concentrated mangroves for evaluation, namely the Tongming Bay zone (reserve 12), the Xinliao–Hean zone (reserve 13), and the Gaoqiao–Anpu zone (reserve 14). The Enping Reserves and Zhenhai Bay Reserve in Guangdong Province (reserve 9) are two adjacent county-level reserves. Since the establishment time of these two reserves was very close and the contiguous area of mangroves at their boundaries was large, the two reserves were considered as a whole for evaluation. The Pinggang Reserves and Haoguang Reserves in Guangdong Province (reserve 10) and the Caiqiao Reserve and Dongchang Reserve in Hainan Province (reserve 20) were also evaluated as a whole for the same reasons. The Mai Po Marshes Reserve (reserve 6) was established by the Hong Kong Special Administrative Region, China, and was added as a Wetland of International Importance under the Ramsar Convention. Due to its high protection level for mangroves, it was classified as a national-level reserve in this study.

2.2. Mangrove Distribution Dataset

The long time-series China’s mangrove distribution dataset was collected from the Natural Resources Satellite Remote Sensing Cloud Service Platform, which included five periods of mangrove distribution data for 1978, 1990, 2000, 2013, and 2018 (the canopy density of the mangrove patches is greater than 20%, and the area is greater than 100 m²) [40,41]. First, the mangrove distribution data for 2018 (MC2018) was obtained using a hybrid method of object-based image analysis, visual interpretation, and field surveys based on 2 m high-resolution satellite images of GF-1 and ZY-3, with an overall accuracy of 99.3% [11]. Specifically, the object-based image analysis mainly included four steps: sample collection, image segmentation, image object classification, and post-processing, which were used for the primary extraction of mangrove plots. Visual interpretation and field surveys were conducted to refine the detected mangrove plots and evaluate the classification accuracy. Then, the MC2018 data were used as the basic reference data, and the mangrove distributions in 2013, 2000, 1990, and 1978 were interpreted using a reverse time order and step-by-step interpretation strategy [42,43]. In addition, considering the low resolution of the historical satellite images and the inability to conduct field surveys to refine the extraction results, three measures were implemented to ensure the classification accuracy: (1) comprehensive interpretation of mangroves combined with multi-tidal inundation information, (2) auxiliary identification of mangroves based on vegetation phenology information, and (3) collaborative interpretation of mangroves based on Google Earth historical high-resolution imagery.

Table 2. The satellite platform and sensors used in mangrove monitoring.

Year	Satellite	Sensor	Resolution
2018	GF-1	PMS	2 m
	ZY-3	NAD and MUX	2 m
2013	GF-1	PMS	2 m
	ZY-3	NAD and MUX	2 m
2000	Landsat 7	ETM+	15 m
1990	Landsat 5	TM	30 m
1978	Landsat 1, 2, and 3	MSS	60 m

presents the details of the satellite data used for interpretation in historical periods.

2.3. Methods

2.3.1. Construction of Evaluation Index System for Landscape Health

In order to comprehensively and accurately evaluate the conservation effectiveness of China’s mangrove reserves, we constructed an evaluation index system for landscape health. First, considering the objective of constructing the evaluation index system and the basic criteria of index selection, i.e., scientific, systematic, operational, and practical, thirteen landscape metrics were selected as evaluation indexes after a review of the literature concerning the application and ecological significance of landscape metrics [32,33,44,45,46]. These thirteen metrics constitute the preliminary evaluation index system, and they can characterize the landscape health status of mangroves in reserves from multiple aspects, including patch area, patch shape, patch distance, habitat fragmentation, and connectivity (

Table 3. Preliminary evaluation index system for landscape health.

Category	Metrics	Ecological Significance	Types
Area	TA	TA is an important index of landscape structure, which can directly reflect the change of mangrove area. In landscape ecological construction, the landscape area is the most important factor used to maintain the stability of the ecosystem.	Positive
	AREA_MN	The mean size of mangrove patches can reflect the fragmentation of the mangrove landscape.	Positive
Shape complexity	SHAPE_MN	SHAPE_MN characterizes the degree of regularity of the patches and the complexity of their edges. When all the patches in the landscape are square, SHAPE_MN is equal to 1. When the shape of the patches deviates from the square, SHAPE_MN increases.	Positive
	SHAPE_AM	SHAPE_AM is an important index for measuring the complexity of landscape spatial patterns and has an impact on many ecological processes.	Positive
	FRAC_MN *	An index to measure the spatial shape complexity of patches with fractal dimension theory. The closer the value tends to 1, the more regular the shape of the patch.	Positive
	FRAC_AM *	An important index to reflect the overall shape characteristics of landscape patterns. It also reflects the impact of human activities on the mangrove landscape pattern to a certain extent.	Positive
Distance	ENN_MN	ENN_MN is widely used to quantify patch isolation and characterize the spatial distribution of discrete or aggregated patches. It also reflects the difficulty of the ecological process of species migration and energy flow.	Negative
Fragmentation	PD	PD is the number of mangrove patches per unit area, which characterizes the degree of fragmentation in the landscape. The higher the patch density, the greater the degree of landscape fragmentation.	Negative
	NP	NP is the number of mangrove patches, the simplest index to measure landscape separation and fragmentation.	Negative
	ED *	ED is used to characterize the length of the edge per unit area, revealing the extent to which the mangrove landscape is divided by the edge.	Negative
Connectivity	CONTIG_MN	CONTIG_MN can measure the proximity between patches of the same type. A large value indicates a high degree of proximity between patches of the same type and good landscape connectivity.	Positive
	COHESION *	COHESION measures the physical connectivity between patches to characterize the habitat connectivity between mangrove patches.	Positive
	AI *	AI characterizes the connectivity between mangrove patches and measures the degree of possible connectivity between mangrove patches.	Positive

* Indexes eliminated by correlation analysis and discrimination analysis. Abbreviations: TA—total area; AREA_MN—mean patch area; SHAPE_MN—mean patch shape index; SHAPE_AM—area-weighted mean patch shape index; FRAC_MN—mean patch fractal dimension; FRAC_AM—area-weighted mean patch fractal dimension; ENN_MN—mean Euclidean nearest neighbor distance distribution; PD—patch density; NP—number of patches; ED—edge density; CONTIG_MN—mean contiguity index distribution; COHESION—patch cohesion index; AI—aggregation index.

).

In addition, according to the ecological significance represented by the indices, these indices were divided into two types: positive index (i.e., the higher the value, the better the landscape health) and negative index (i.e., the lower the value, the better the landscape health). For example, the complexity of patch shape can reflect the impact of human activities (e.g., population migration, urban expansion, and aquaculture) on mangroves, and human activities generally result in the simplification of patch shape [47,48]. Therefore, the increase in the patch shape complexity (a higher value) can indicate the effectiveness of mangrove conservation. The fragmentation of mangrove habitats may lead to a series of negative impacts on mangrove ecosystems, such as weakened stability, reduced service capacity, and reduced biodiversity [49,50]. Thus, the reduction of patch fragmentation (a lower value) can also reflect the positive conservation effect of the reserve on mangroves. Finally, since the inconsistent spatial resolution of the images used for interpretation in different periods, to prevent the uncertainty of the final evaluation results caused by scale effect, the shapefiles of five periods of mangrove distribution data were projected into 10 m raster data [51]. The spatial resolution was set to 10 m (i.e., a raster size was 100 m²) because the minimum patch size of mangroves extracted during the 2018 mapping procedure was 100 m². Then, the selected landscape metrics were calculated using FRAGSTATS v4.2.1 software [52].

The reduction of landscape metrics in the preliminary evaluation index system was necessary to eliminate information redundancy, confusion, and duplication that may exist among them. In this study, a combination of correlation analysis and discrimination analysis was used to screen and optimize the preliminary evaluation index system by eliminating redundant landscape metrics [32,35]. First, a correlation analysis was carried out based on the Pearson correlation coefficient. If the correlation coefficient of a pair of metrics was greater than 0.9, a high correlation was considered to exist, and only one of the two metrics was retained. Following the correlation analysis, a discrimination analysis was carried out based on the coefficient of variation. The coefficient of variation can eliminate the influence of dimensional differences, and the degree of data dispersion between indices of different demission can be compared. The higher the value of the coefficient of variation, the greater the internal dispersion of the landscape index, i.e., the better the discrimination of the landscape index. If the coefficient of variation was less than 0.1, the landscape metric was considered to be poorly distinguishable, and it was eliminated from the preliminary evaluation index system. The formula of the coefficient of variation is as follows:

$${\mathrm{c}}_{\mathrm{v}}=\frac{{\mathsf{\sigma }}_{\mathrm{j}}}{{\mathsf{\mu }}_{\mathrm{j}}}$$

(1)

where ${\mathrm{c}}_{\mathrm{v}}$ is the coefficient of variation, ${\mathsf{\sigma }}_{\mathrm{j}}$ is the standard deviation of index $\mathrm{j}$, and ${\mathsf{\mu }}_{\mathrm{j}}$ is the mean value of index $\mathrm{j}$.

The calculation results of the correlation coefficient and the coefficient of variation are given in Table S1. As a result, eight landscape metrics were retained to construct the final evaluation index system: TA, AREA_MN, SHAPE_MN, SHAPE_AM, PD, NP, ENN_MN, and CONTIG_MN.

2.3.2. Establishment and Calculation of the Landscape Health Composite Index

Although the eight representative landscape metrics in the final evaluation index system can reflect the landscape pattern from different aspects, the landscape health status of mangroves is not dependent on a single index but is influenced by them collectively. In order to comprehensively describe the landscape health status of mangroves, a landscape health composite index (LHCI) was established based on the entropy weight method. The entropy weight method is a commonly used weighting method to determine the weight of indices in a system based on information entropy by comprehensively measuring the amount of information provided by each index [53,54,55]. Unlike the subjective weighting method (e.g., Delphi method, analytic hierarchy process method), the entropy weight method can effectively avoid the distortion of weights caused by the interference of personal factors, ensuring the credibility and objectivity of the weight values. The primary process is as follows:

(1)

Constructing the original evaluation index matrix X:

$$\mathrm{X}={\left({\mathrm{x}}_{\mathrm{ij}}\right)}_{\mathrm{m}\times \mathrm{n}}$$

(2)

where m is the number of reserves, n is the number of evaluation indexes, and ${\mathrm{x}}_{\mathrm{ij}}$ is the $\mathrm{j}\mathrm{th}$ index value of the $\mathrm{i}\mathrm{th}$ reserve.

(2)

Performing data normalization to obtain the normalization matrix $\mathrm{X}\prime$. Data standardization can eliminate the differences among indices caused by the inconsistency of the dimension and direction. For positive index ${\mathrm{x}}_{\mathrm{ij}}$:

$${\mathrm{x}}_{\mathrm{ij}}^{\prime }=\frac{{\mathrm{x}}_{\mathrm{ij}}-\mathrm{MIN}\left({\mathrm{x}}_{\mathrm{j}}\right)}{\mathrm{MAX}\left({\mathrm{x}}_{\mathrm{j}}\right)-\mathrm{MIN}\left({\mathrm{x}}_{\mathrm{j}}\right)}$$

(3)

For negative index ${\mathrm{x}}_{\mathrm{ij}}$:

$${\mathrm{x}}_{\mathrm{ij}}^{\prime }=\frac{\mathrm{MAX}\left({\mathrm{x}}_{\mathrm{j}}\right)-{\mathrm{x}}_{\mathrm{ij}}}{\mathrm{MAX}\left({\mathrm{x}}_{\mathrm{j}}\right)-\mathrm{MIN}\left({\mathrm{x}}_{\mathrm{j}}\right)}$$

(4)

where ${\mathrm{x}}_{\mathrm{ij}}^{\prime }$ is the normalized value of ${\mathrm{x}}_{\mathrm{ij}}$, $\mathrm{MAX}\left({\mathrm{x}}_{\mathrm{j}}\right)$ is the maximum value of the $\mathrm{j}\mathrm{th}$ index, $\mathrm{MIN}\left({\mathrm{x}}_{\mathrm{j}}\right)$ is the minimum value of the $\mathrm{j}\mathrm{th}$ index. Then, the normalization matrix $\mathrm{X}\prime$ is obtained:

$$\mathrm{X}\prime ={({\mathrm{x}}_{\mathrm{ij}}^{\prime })}_{\mathrm{m}\times \mathrm{n}}$$

(5)

(3)

Calculating the proportion value (${\mathrm{p}}_{\mathrm{ij}}$) of the $\mathrm{i}\mathrm{th}$ reserve under the $\mathrm{j}\mathrm{th}$ index:

$${\mathrm{p}}_{\mathrm{ij}}=\frac{{\mathrm{x}}_{\mathrm{ij}}^{\prime }}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{ij}}^{\prime }}$$

(6)

(4)

Calculating the entropy value (${\mathrm{e}}_{\mathrm{j}}$) of the $\mathrm{j}\mathrm{th}$ index:

$${\mathrm{e}}_{\mathrm{j}}=\frac{1}{\ln \mathrm{m}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{p}}_{\mathrm{ij}}\ln {\mathrm{p}}_{\mathrm{ij}}$$

(7)

(5)

Calculating the difference coefficient value (${\mathrm{g}}_{\mathrm{j}}$) of the $\mathrm{j}\mathrm{th}$ index:

$${\mathrm{g}}_{\mathrm{j}}=1-{\mathrm{e}}_{\mathrm{j}}$$

(8)

(6)

Calculating the weight value (${\mathrm{w}}_{\mathrm{j}}$) of the $\mathrm{j}\mathrm{th}$ index:

$${\mathrm{w}}_{\mathrm{j}}=\frac{{\mathrm{g}}_{\mathrm{j}}}{{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{g}}_{\mathrm{j}}}$$

(9)

(7)

Calculating the landscape health composite index value (${\mathrm{z}}_{\mathrm{i}}$) of the $\mathrm{i}\mathrm{th}$ reserve:

$${\mathrm{z}}_{\mathrm{i}}={{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{ij}}^{\prime }$$

(10)

2.3.3. Assessment of Mangrove Dynamics

The dynamic change of mangrove areas is an objective and direct reflection of the conservation effectiveness and the evolution of mangrove landscape patterns. To objectively analyze and evaluate the dynamic degree of mangroves, we calculated the annual land change rate (ALCR) of mangroves [56]. The formula of ALCR is as follows:

$$\mathrm{ALCR}=\frac{{\mathrm{S}}_{\mathrm{a}}-{\mathrm{S}}_{\mathrm{b}}}{{\mathrm{S}}_{\mathrm{b}}}\times \frac{1}{\mathrm{T}}\times 100\%$$

(11)

where ${\mathrm{S}}_{\mathrm{b}}$ represent the area of mangroves at the time point before the reserve was established. ${\mathrm{S}}_{\mathrm{a}}$ represent the area of mangroves at the end of the study period, i.e., 2018. T is the number of years from the time point before the establishment of the reserve to the time point at the end of the study period.

2.3.4. Evaluation of Mangrove Conservation Effectiveness

In this study, we chose four indicators to analyze the dynamic changes of mangrove landscape health status for the period before and after the establishment of reserves: the amplitude of LHCI ($\mathrm{C}$), the amplitude ratio of LHCI (${\mathrm{C}}_{\mathrm{r}}$), the slope of LHCI (${\mathrm{k}}_1$), and the slope of the amplitude ration of LHCI (${\mathrm{k}}_2$) [31]. The conservation effectiveness of mangrove reserves in China was evaluated based on the analysis results.

Since the establishment time of each reserve varied greatly (the earliest was 1975, the latest was 2005), we defined the time point before protection (${\mathrm{T}}_{\mathrm{b}}$) (i.e., the time point before the reserve was established) by taking into account the establishment time of each reserve and the time point contained in the mangrove distribution dataset. The definition of the time point before and after protection is shown in

Table 4. Definition of the time point of reserves before and after protection.

Range of Establishment Time of Reserves 1	Time Point $\mathbf{before} \mathbf{Protection} \left({\mathbf{T}}_{\mathbf{b}}\right)$	Time Point $\mathbf{after} \mathbf{Protection} \left({\mathbf{T}}_{\mathbf{a}}\right)$
1978–1989	1978	2018
1990–1999	1990	2018
2000–2005	2000	2018

¹ Excluding the Ximen Island reserve and the Mai Po reserve.

. In addition, the mangroves in the Ximen Island reserve were artificially introduced and planted, and there were no natural mangroves before the establishment of the reserve (i.e., 2005). Therefore, its ${\mathrm{T}}_{\mathrm{b}}$ was defined as the year closest to the establishment time of the reserve (i.e., 2013). The Mai Po reserve was established in 1975, earlier than the start time of the mangrove distribution dataset. Therefore, its ${\mathrm{T}}_{\mathrm{b}}$ was also defined as the year closest to the establishment time of the reserve (i.e., 1978). Finally, the time point after the protection (${\mathrm{T}}_{\mathrm{a}}$) of each reserve was defined as 2018.

The amplitude of LHCI ($\mathrm{C}$) is the difference between the value of LHCI at T_a and the value of LHCI at T_b, i.e., the net increase or decrease in the LHCI. A positive value indicates the improvement of mangrove landscape health status after protection, and a negative value indicates the deterioration of mangrove landscape health status after protection. The amplitude ratio of LHCI (${\mathrm{C}}_{\mathrm{r}}$) is the ratio of the $\mathrm{C}$ to the value of LHCI at T_b, which characterizes the significant degree of the improvement or deterioration of mangrove landscape health status after protection. The slope of LHCI (${\mathrm{k}}_1$) is the slope of the line fitted by the value of LHCI at each time point from T_b to T_a, which characterizes the changing trend of the LHCI after protection; a positive value indicates an increasing trend of LHCI, and a negative value indicates a decreasing trend of LHCI. The formula of $\mathrm{C}$ and ${\mathrm{C}}_{\mathrm{r}}$ are as follows:

$$\mathrm{C}={\mathrm{Z}}_{\mathrm{a}}-{\mathrm{Z}}_{\mathrm{b}}$$

(12)

$${\mathrm{C}}_{\mathrm{r}}=\frac{\mathrm{C}}{{\mathrm{Z}}_{\mathrm{b}}}\times 100\%$$

(13)

where ${\mathrm{Z}}_{\mathrm{b}}$ and ${\mathrm{Z}}_{\mathrm{a}}$ are the value of LHCI at ${\mathrm{T}}_{\mathrm{b}}$ and ${\mathrm{T}}_{\mathrm{a}}$, respectively.

In addition, although some reserves have not significantly improved the landscape health status of mangroves in the short term after protection (i.e., $\mathrm{C}<0$), they have effectively mitigated the deterioration trend of landscape health status, and their conservation effects should also be recognized. Therefore, the amplitude ratio of LHCI for each period from ${\mathrm{T}}_{\mathrm{b}}$ to ${\mathrm{T}}_{\mathrm{a}}$ (${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$) of the reserve was also calculated. The formula of ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$ is as follows:

$${\mathrm{C}}_{\mathrm{i}}={\mathrm{Z}}_{\mathrm{m}}^{\mathrm{i}}-{\mathrm{Z}}_{\mathrm{n}}^{\mathrm{i}}$$

(14)

$${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}=\frac{{\mathrm{C}}_{\mathrm{i}}}{{\mathrm{Z}}_{\mathrm{n}}^{\mathrm{i}}}\times 100\%$$

(15)

where ${\mathrm{Z}}_{\mathrm{n}}^{\mathrm{i}}$ and ${\mathrm{Z}}_{\mathrm{m}}^{\mathrm{i}}$ are the value of LHCI at the beginning and end of the $\mathrm{i}\mathrm{th}$ period from ${\mathrm{T}}_{\mathrm{b}}$ to ${\mathrm{T}}_{\mathrm{a}}$. Finally, the slope of the amplitude ratio of LHCI (${\mathrm{k}}_2$) is the slope of the line fitted by the value of ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$, which characterizes the changing trend of the decreased degree of LHCI after protection. A positive value indicates that the decreased degree of LHCI is weakening, and a negative value indicates that the decreased degree of LHCI is still increasing.

3. Results

3.1. Dynamics of Mangrove Area after Protection

The net change area and ALCR of mangroves in each reserve are shown in

Table 5. Dynamic changes of mangrove area in reserves before and after protection.

Reserve No.	Reserve	${\mathbf{S}}_{\mathbf{b}} \left(\mathbf{ha}\right)$	${\mathbf{S}}_{\mathbf{a}} \left(\mathbf{ha}\right)$	Net Change (ha)	$\mathbf{K}$
1	Ximen Island	29.49	36.28	6.79	4.60%
2	Quanzhou Bay	161.55	290.17	128.62	4.42%
3	Jiulong Estuary	98.09	404.85	306.76	7.82%
4	Zhangjiang Estuary	49.73	86.40	36.67	2.63%
5	Danshui Estuary	61.12	151.95	90.83	3.72%
6	Mai Po	173.77	535.84	362.07	5.21%
7	Futian	47.00	139.94	92.94	4.94%
8	Qi’ao Island	22.37	525.48	503.11	50.82%
9	Enping and Zhenhai Bay	484.17	904.69	420.52	4.83%
10	Pinggang and Haoguang	603.20	769.97	166.77	1.54%
11	Dianbai	304.19	313.33	9.14	0.11%
12	Tongming Bay	4741.50	2207.83	−2533.69	−1.91%
13	Xinliao–Hean	544.87	735.19	190.32	1.25%
14	Gaoqiao–Anpu	959.46	1472.45	512.99	1.91%
15	Shankou	1201.70	1973.04	771.31	2.29%
16	Maowei Gulf	898.74	2162.70	1263.96	7.81%
17	Beilun Estuary	814.71	1039.33	224.62	0.98%
18	Dongzhaigang	1605.30	1841.45	236.15	0.37%
19	Huachang Bay	319.92	250.91	−69.01	−0.77%
20	Caiqiao and Dongchang	229.52	284.00	54.48	0.59%
21	Xinying Bay	906.23	508.36	−397.87	−1.57%
22	Sanya River	33.32	34.28	0.96	0.10%
23	Qinglan	1715.30	792.51	−922.82	−1.34%

. In general, although the area of mangroves in a few reserves still decreased to a certain extent after the reserves were established, most reserves showed an increasing trend in the mangrove area. Specifically, the mangrove area in 83% of the reserves increased after protection, among which the most significant increase in mangrove area is the Qi’ao Island (reserve 8), with an ALCR of 50.82% and a net increase of 503.11 ha. In addition, the area of mangroves in Jiulong Estuary (reserve 3) and Maowei Gulf (reserve 16) also increased notably, with an ALCR of 7.82% and 7.81%, respectively. However, the mangrove area did not increase in the four reserves after protection: Qinglan (reserve 23), Huachang Bay (reserve 19), Xinying Bay (reserve 21), and Tongming Bay (reserve 12), which were mainly concentrated in the northern part of Hainan Province. Notably, the mangrove area changes in different zones of the Zhanjiang reserve varied considerably after the reserve was established. The mangrove area in Xinliao–Hean (reserve 13) and Gaoqiao–Anpu (reserve 14) both increased, while the mangrove area in Tongming Bay (reserve 12) decreased significantly.

The statistics of mangrove area changes in the reserves of different levels and districts before and after protection are shown in

Table 6. Dynamic changes of mangrove area in the reserves of different levels before and after protection.

Level	${\mathbf{S}}_{\mathbf{b}} \left(\mathbf{ha}\right)$	${\mathbf{S}}_{\mathbf{a}} \left(\mathbf{ha}\right)$	Net Change (ha)
National	10,167.58	10,067.75	−99.83
Provincial	2957.20	4327.66	1370.46
Municipal and county	2880.55	3065.54	184.99
Total	16,005.33	17,460.95	1455.62

and

Table 7. Dynamic changes of mangrove area in the reserves of different districts before and after protection.

District	${\mathbf{S}}_{\mathbf{b}} \left(\mathbf{ha}\right)$	${\mathbf{S}}_{\mathbf{a}} \left(\mathbf{ha}\right)$	Net Change (ha)
Zhejiang	29.49	36.28	6.79
Fujian	309.37	781.42	472.05
Taiwan	61.12	151.95	90.83
Guangdong	7706.78	7068.88	−637.90
Hong Kong	173.77	535.84	362.07
Guangxi	2915.18	5175.07	2259.89
Hainan	4809.62	3711.51	−1098.11
Total	16,005.33	17,460.95	1455.62

. The total mangrove area in reserves increased from 16,005.33 ha to 17,460.95 ha after protection, with a net increase of 1455.62 ha. From the perspective of mangrove area changes in the reserves of different levels, the increase in mangrove area in provincial-level, municipal-level, and county-level reserves was higher than that in national-level reserves. The most significant change of mangrove area was in the provincial-level reserves, which increased by 1370.46 ha. In contrast, the change of mangrove area in national-level reserves was not significant. From the perspective of mangrove area changes in the reserves of different districts, the most significant increase in mangrove area after protection was in Guangxi Province, where the total mangrove area of reserves increased by 2259.89 ha, followed by Fujian and Hong Kong. In addition, the total mangrove area in the reserves of Taiwan and Zhejiang Province also increased, while the total mangrove area in the reserves of Hainan and Guangdong Province still decreased.

The ALCR and the dynamics of mangrove areas in reserves after protection can characterize the conservation effectiveness to a certain extent. However, area indicators alone cannot fully reflect the changes in mangrove landscape quality or landscape health status. For example, mangrove forests that experienced severe damage can rapidly increase their area by artificial planting. However, problems such as the aggravation of landscape fragmentation and poorer habitat connectivity caused by external damage are difficult to recover quickly in a short period. Therefore, it is one-sided and incomprehensive to evaluate the conservation effectiveness of mangroves based only on the area indicator.

3.2. Conservation Effectiveness of Mangrove Reserves in China

3.2.1. Analysis of Conservation Effectiveness in Mangrove Reserves

The establishment of China’s mangrove reserves focused on factors including the spatial distribution characteristics of mangroves and the “rescuing protection” of mangroves. Therefore, taking the establishment time of the reserve as the node, we comprehensively evaluated the reserves’ conservation effectiveness on mangroves based on the dynamics of the mangrove landscape health status for the period before and after the establishment of the reserves. The dynamics of mangrove landscape health status in each reserve can be divided into the following four patterns.

(1)

$\mathrm{C}>0$ and ${\mathrm{k}}_1>0$, representing that the LHCI was net increased after protection and showed a continuously increasing trend. In this pattern, the landscape health status of mangroves improved significantly compared to that before the reserve was established and maintained a constant positive trend, indicating that the reserves achieved remarkable conservation effectiveness on mangroves. There are ten reserves (reserves 2, 3, 8, 9, 10, 14, 15, 16, 18, and 22) that conformed to this pattern. Among them, the C_r of the Qi’ao Island (reserve 8), Maowei Gulf (reserve 16), and Enping (reserve 9) reached 519.78%, 45.01%, and 25.65%, respectively, which were the three most outstanding reserves in terms of conservation effectiveness.

(2)

$\mathrm{C}>0$ and ${\mathrm{k}}_1<0$, representing that the LHCI was net increased after protection but showed a decreasing trend. In this pattern, the landscape health status of mangroves improved compared to that before the reserve was established, indicating that the reserves achieved good conservation effectiveness on mangroves. However, the landscape health status of mangroves did not maintain a positive trend, i.e., the LHCI only increased greatly in isolated periods and then decreased. The reserve that conformed to this pattern was Caiqiao (reserve 20), which needs attention.

(3)

$\mathrm{C}<0$ and ${\mathrm{k}}_2>0$, representing that the LHCI was decreased after protection, but the decreasing trend continuously slowed down. In this pattern, the landscape health status of mangroves gradually turned better after the reserve was established, indicating that the reserves achieved a certain protective effect while the conservation effectiveness was relatively insignificant. There are seven reserves (reserves 4, 7, 11, 12, 19, 21, and 23) that conformed to this pattern. Among them, the latest ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$ in Zhangjiang Estuary (reserve 4), Futian (reserve 7), Tongming Bay (reserve 12), and Huachang Bay (reserve 19) has turned from negative to positive in recent periods. The protection and restoration of mangroves in these reserves should be further enhanced to accelerate the improvement of the landscape health status of mangroves.

(4)

$\mathrm{C}<0$ and ${\mathrm{k}}_2<0$, representing that the LHCI was still decreased after protection, and the decreasing trend did not slow down. In this pattern, the landscape health status of mangroves did not improve obviously compared to that before the reserve was established, indicating that the conservation effectiveness on mangroves was ordinary. Five reserves (reserves 1, 5, 6, 13, and 17) conformed to this pattern, and special attention and focus to these reserves needed to be strengthened.

Based on the four patterns of the dynamics of mangrove landscape health status in reserves, conservation effectiveness can be categorized into three levels: excellent, good, and ordinary (

Table 8. Criteria and results of conservation effectiveness evaluation of reserves.

Conservation Effectiveness	$\mathbf{C}$	${\mathbf{k}}_1$	${\mathbf{k}}_2$	Number	Percent
Excellent	$\mathrm{C}>0$	${\mathrm{k}}_1>0$	/	10	43%
Good	$\mathrm{C}>0$	${\mathrm{k}}_1<0$	/	8	35%
	$\mathrm{C}<0$	/	${\mathrm{k}}_2>0$
Ordinary	$\mathrm{C}<0$	/	${\mathrm{k}}_2<0$	5	22%

). The results show that mangrove reserves in China achieved outstanding conservation effectiveness, with 43% of the reserves significantly improving the landscape health status of mangroves, 35% of the reserves maintaining good mangrove landscape health status, and only 22% of the reserves having ordinary protection efficacy. The detailed evaluation results of each reserve are shown in Table S2.

3.2.2. Comparison of Conservation Effectiveness in Different Levels of Reserves

The statistics of reserves’ evaluation results based on protection level are shown in

Table 9. Statistics of evaluation results of reserves based on protection levels.

Level	Conservation Effectiveness	Number	Percent
National	Excellent	3	30%
	Good	3	30%
	Ordinary	4	40%
Provincial	Excellent	4	66%
	Good	1	17%
	Ordinary	1	17%
Municipal and county	Excellent	4	57%
	Good	3	43%
	Ordinary	0	0%

. The results showed that 60% of the national-level reserves, 83% of the provincial-level reserves, and all the municipal- and county-level reserves were excellent and good reserves, which showed remarkable conservation effectiveness of mangrove reserves in China. In addition, the Gaoqiao–Anpu (reserve 14) at the national level, the Qi’ao Island (reserve 8) at the provincial level, and the Enping (reserve 9) at the county level were the most effective in protecting mangroves among each level of reserves, with the ${\mathrm{C}}_{\mathrm{r}}$ of 12.13%, 519.78%, and 25.65%, respectively. In general, the conservation effectiveness of provincial-, municipal-, and county-level reserves was more significant than the national-level reserves, with excellent reserves accounting for 66% and 57%, respectively.

3.2.3. Comparison of Conservation Effectiveness in Different Districts of Reserves

From the evaluation results of mangrove reserves in each province (Figure 2), the reserves in Guangxi, Guangdong, and Fujian Provinces are the most effective in protecting mangroves, with the proportion of excellent reserves at 50% and above. From the changes of the LHCI before and after the establishment of reserves in different provinces (Figure 3), the LHCI increased the most in Guangxi and Guangdong Provinces, followed by Fujian Province, which indicated that the reserves in Guangxi and Guangdong Provinces have larger conservation effectiveness than those in other provinces. There is a clear outlier in Guangdong Province (Figure 3), which represents the Qi’ao Island, the earliest reserve where mangrove planting was carried out in China. Before protection, the local mangroves were severely destroyed by human activities (e.g., urbanization). After establishing the reserve, the government invested special funds to carry out mangrove species introduction and mangrove reforestation projects. The landscape health status of mangroves has been significantly improved. Therefore, the LHCI of the reserve showed a dramatic improvement accordingly.

Considering the specificity of the spatial distribution of mangrove forests in the Zhanjiang reserve, we divided it into three zones with concentrated mangroves for evaluation. The results show that the effectiveness of conservation on mangroves in Gaoqiao–Anpu, Tongming Bay, and Xinliao–Hean was excellent, good, and ordinary, respectively. Given the obvious zonal differences in the evaluation results of the Zhanjiang reserve, we take the Zhanjiang reserve as an example to conduct an in-depth analysis of the temporal and spatial evolution of the mangrove landscape pattern in each zone to verify the reliability of the evaluation results.

Figure 4a shows the changes in mangrove patches in the Tongming Bay zone. The mangrove area in this zone continued to decrease after the establishment of the reserve. The most severe mangrove area retreat occurred between 1990 and 2013, with high fragmentation of mangrove patches and a significant decrease in spatial aggregation and habitat connectivity of patches. However, there were no significant changes in the mangrove patches from 2013 to 2018, and the landscape health status of mangroves remained stable. From the changes of the corresponding LHCI (Figure 5), the LHCI continued to decline from 1990 to 2013, with a slight increase from 2013 to 2018; 2013 was the turning point when the mangrove landscape health status changed. Although the current landscape health status has not been significantly improved, the changes in the LHCI and ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$ show that the deteriorating trend of landscape health status has been completely controlled, and the landscape health status has gradually turned better. Therefore, the zone has achieved a certain protective effect. For the Xinliao–He’an zone (Figure 4b), the mangrove area increased significantly in the early stage from 1990 to 2000, and some new small patches appeared around the original mangrove patches. However, after 2000, many of the original large patches were obviously destroyed and tended to be fragmented. The changes in the LHCI and ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$ (Figure 5) show that the LHCI only slightly improved between 1990 and 2000. After 2000, the LHCI continued to decline, and the decreasing trend has not been completely controlled. Therefore, the conservation effectiveness of this zone is not apparent, and it is classified into ordinary grades. The changes of mangrove patches in the Gaoqiao–Anpu zone (Figure 4c) show that the spatial distribution of mangroves has been continuously expanding after the establishment of the reserve, and the original small patches have gradually developed into large patches with high spatial aggregation. Moreover, the shape of the patches tends to be complex, and the habitat connectivity between patches is obviously enhanced. The changes in the LHCI (Figure 5) show that the LHCI throughout the entire period and has been dramatically improved compared to that before the reserve was established, indicating significant conservation effectiveness on mangroves.

The above analysis shows that the changes in mangrove landscape patterns are consistent with the situation reflected by LHCI and ${\mathrm{C}}_{\mathrm{r}}^{\mathrm{i}}$. The four indicators ($\mathrm{C}$, ${\mathrm{C}}_{\mathrm{r}}$, ${\mathrm{k}}_1$, and ${\mathrm{k}}_2$) we selected can effectively describe the dynamics of mangroves landscape health status in reserve. Therefore, the evaluation result based on the LHCI is accurate and reliable. In addition, compared with the evaluation of the Zhanjiang reserve as a whole, the results obtained by the zoning evaluation in this study are more reasonable. Differentiated protection strategies can be adopted according to the spatial differentiation characteristics of its conservation effectiveness, focusing on strengthening the protection and restoration of mangroves in the Tongming Bay zone and Xinliao–He’an zone.

4. Discussion

As the ecological and socio-economic functions of mangroves are increasingly valued in China, the government has issued a series of laws and regulations to protect mangroves. The number of mangrove reserves is increasing, and about 64% of mangroves have been protected [11]. However, the quantitative enhancement of mangrove reserves does not always guarantee qualitative improvement [57]. The conservation effectiveness evaluation is necessary to enhance the conservation level of mangrove reserves and promote the restoration of mangrove resources.

Previous studies based on remote sensing mainly focused on the analysis and evaluation of the conservation effects of mangrove reserves from the perspective of spatial-temporal changes in mangroves, such as the research carried out in the Zhanjiang reserve [58], Qinglan reserve [59], Quanzhou Bay reserve [56], and Zhangjiang Estuary [60]. There is no doubt that the increase in the area of mangroves can indicate the conservation effects of reserves. The maintenance of mangrove cover is also an essential guarantee for preserving the integrity and biodiversity of mangrove ecosystems [61]. However, the mangrove area is only one indicator to evaluate conservation effectiveness [29]. If there is insufficient knowledge of the quality and spatial arrangement of mangrove habitat, the area indicator alone is not adequate to guide the implementation of conservation actions. Though measuring and understanding changes in mangroves is an essential step in developing conservation policies and identifying conservation priorities, other indicators of habitat health, such as connectivity and fragmentation, are also critical indicators for conservation effectiveness evaluation [49,62]. Therefore, it is more comprehensive and reliable to evaluate the conservation effectiveness from the perspective of landscape pattern change instead of simply considering the area change.

Different from previous studies, our analysis of changes in landscape metrics was not conducted independently [63,64,65]. Independent analysis of landscape metrics is feasible for evaluating the conservation effectiveness of a few reserves [34]. When the number of reserves to be evaluated is large, or the changes in landscape metrics are complex, it may be difficult to judge which reserve is more effective in protecting mangroves. For example, AREA_MN and PD may increase simultaneously (by different amplitudes) over a certain period. However, as pointed out in Section 2.3.1, one is a positive index, and the other is a negative index. Therefore, we comprehensively considered the ecological significance of different landscape metrics by establishing the LHCI. This measure is reasonable because one or a class of landscape metrics only characterizes one aspect of the landscape pattern. The detailed analysis of the temporal and spatial evolution of the mangrove landscape pattern in three zones of the Zhanjiang reserve also verified the reliability of the evaluation method based on the LHCI. On the other hand, as far as the research scale is concerned, there is still a lack of specialized research on the conservation effectiveness evaluation of mangrove reserves at the national scale in China. Zheng et al. [66] and Jia et al. [28] have conducted a national-scale evaluation of the conservation effectiveness of mangrove reserves, which only included reserves at the national level. Zheng et al. [66] evaluated the conservation effectiveness of China’s national wetland reserves from three aspects: conservation value, wetland changes, and functional zoning adjustment. Although the evaluation index system in their study was comprehensive, the difficulty of collecting reserves database (e.g., environmental data, protection value data, species data) limited the promotion and application of the method. Jia et al. [28] analyzed the conservation effectiveness of national mangrove reserves based on the changes in mangrove forests between 1973 and 2015. However, as only the area was considered, the information provided by the research results was not exhaustive.

In this study, the research objects covered mangrove reserves at all levels (national, provincial, municipal, and county levels), and the study area covered the main distribution areas of mangroves in China (six coastal provinces and one special administrative region). The comprehensive evaluation results obtained from the perspective of landscape health can provide a practical reference for the construction and management of mangrove reserves and the formulation and adjustment of protection policies. For those good reserves, we suggest that long-term protection plans should be formulated, and more reasonable and targeted protection measures should be developed to consolidate and improve conservation effectiveness, such as conducting field surveys of critical mangrove habitats and increasing the frequency of remote sensing monitoring. For those ordinary reserves, we suggest that more investment should be increased to improve conservation effectiveness as soon as possible, carrying out mangrove species introduction projects and implementing replantation and afforestation programs.

However, there are still limitations in remote sensing-based conservation evaluation. First, a time-series mangrove distribution dataset is necessary for remote sensing-based evaluation methods, but it is difficult to produce unbiased maps due to differences in the raw images used in the mapping process (e.g., satellite sensors, image resolutions). Although the dataset can be converted to a consistent spatial resolution, the conversion process can also generate errors. In addition, the change of spatial resolution (pixel size) often affects landscape metrics. On the other hand, large-scale mangrove afforestation is a primary conservation measure in China [67]. For example, in May 2017, the State Forestry Administration and the National Development and Reform Commission released the Planning of Constructing National Coastal Shelter Forests (2016–2025). In this plan, mangrove reforestation was listed as a critical project, aiming to plant 48,650 ha mangroves in 62 counties. Nevertheless, mangroves planted in most restoration projects usually consisted of only a single species (e.g., Sonneratia apetala), which may reduce regional biodiversity. Previous studies have demonstrated the great potential of mapping the spatial distribution of mangrove species based on satellite imagery [68,69,70]. Future research should give some consideration to species diversity to further improve the evaluation of conservation effectiveness based on remote sensing.

5. Conclusions

This study constructed an evaluation index system for landscape health and proposed a landscape health composite index (LHCI) to characterize the landscape health status of mangroves. Using a long time-series China’s mangrove distribution dataset in the past 40 years, the conservation effectiveness of mangrove reserves in China was evaluated by analyzing the dynamics of mangrove area and mangrove landscape health status before and after the establishment of the reserves.

The dynamics of mangrove areas show that the mangrove area in 83% of the reserves increased after the establishment of reserves. In addition, the increase in mangrove area in provincial-level, municipal-level, and county-level reserves was higher than that in national-level reserves, and the most significant increase in mangrove area was in Guangxi, followed by Fujian and Hong Kong. The evaluation results show that reserves protected mangroves effectively in China, with 43% of the reserves significantly improving the landscape health status of mangroves and 35% of the reserves maintaining good mangrove landscape health status. The most effective reserves protecting mangroves were the Qi’ao Island reserve, Maowei Gulf reserve, and Enping reserve. Moreover, the conservation effectiveness of provincial-, municipal- and county-level reserves was more significant than the national-level reserves, with excellent reserves accounting for 66% and 57%, respectively. The reserves in Guangxi, Guangdong, and Fujian Provinces showed more significant protection effects, with the proportion of excellent reserves at 50% and above. Although there are limitations, this study nonetheless provides a sound evaluation to understand the conservation effectiveness of mangrove reserves, which can also be used as a guide for conservation policymaking and scientific management of reserves.