Assessing the ecological vulnerability of protected areas by using Big Earth Data

ABSTRACT

The ecological vulnerability of global protected areas (PAs) is linked to the 2030 United Nations Sustainable Development Goals. We developed a novel degree of ecological vulnerability (DEV) index based on Big Earth Data to assess the ecological vulnerability of PAs. This is a transparent, repeatable, large-scale and rapid assessment method. We applied the method to case studies of International Union for Conservation of Nature (IUCN) PAs in the basins of the Aral Sea, the Caspian Sea, the Black Sea and the Lake Baikal. Between 2001 and 2015, the spatial pattern of ecological vulnerability in the study area was relatively stable. The Caspian Sea basin and protected landscape/seascape (category V) showed high ecological vulnerability, while the Black Sea basin and wilderness areas (category I_b) showed low ecological vulnerability. Big Earth Data shows great vitality in PAs ecological vulnerability assessment. Strengthening international cooperation is an important means to break the bottleneck of ecological environment and resource endowment in these important international basins.

KEYWORDS:

• Protected areas
• ecological vulnerability
• assessment
• remote sensing
• sustainable development goals

1. Introduction

Global biodiversity is declining faster than at any time in human history, which has serious implications for human wellbeing (Pimm et al. 2014; Pimm, Jenkins, and Li 2018; Di Marco et al. 2019; Dinerstein et al. 2019). Aichi target 11 sets a goal of achieving a global area of 17% terrestrial PA coverage and 10% marine PA coverage by 2020 (Visconti et al. 2019; Leendertz and Kalema-Zikusoka 2021). However, the ongoing global assessments indicated that the vast majority of these targets have not been met, and the situation of some targets has worsened (Butchart, Di Marco, and Watson 2016; Venter et al. 2018; Santini et al. 2019; Woodley et al. 2019; Maxwell et al. 2020). The post-2020 global target for PAs will be discussed at the 15th meeting of the Biodiversity Convention in 2021 in China (Yang et al. 2020; UNEP 2020), and there are suggestions that we need a ‘Global Safety Net’ to reverse biodiversity loss and stabilize the Earth’s climate (Dinerstein et al. 2020; Hanson et al. 2020). Assessing PA effectiveness is remarkably challenging, whereas there is overwhelming evidence that the means currently allocated to PAs are insufficient, it remains unclear the extent to which these conservation tools are already contributing to the long-term conservation of nature, individually as well as globally (IUCN 2018; Loiseau et al. 2020; Rodrigues and Cazalis 2020; Xu et al. 2020).

Ecological vulnerability is a topic of growing concern and interest in the field of global environmental change and sustainable development research (Jiang et al. 2018; Xu et al. 2016). In the past decade, several approaches and methods have been proposed to quantitatively evaluate ecological vulnerability (Fuentes, Limpus, and Hamann 2011; Zhang et al. 2017; Thiault et al. 2018). Although these methods differ slightly depending on the target of the application, almost all consider the same key parameters: sensitivity, exposure and adaptive capacity (McCarthy et al. 2001; Cuevas, Liceaga-Correa, and Uribe-Martínez 2019). The published methods have employed different quantitative techniques, taking into account the fact that evaluation indexes cannot reflect the quality of the ecosystem in a linear way, both with and without explicit spatial modules (Holling 1973; Hong et al. 2016). As a transparent, repeatable, large scale and rapid method, Big Earth Data may provide the methodologies to generate essential knowledge from diverse and complex data sources to ensure a sustainable human society that is essential for the preservation of the planet (Stephenson et al. 2017; Guo 2018; Guo et al. 2020; Zheng et al. 2012; Zheng 2020). Given the definition of ecological vulnerability (Turner et al. 2003), the conceptual model of vulnerability reflects the degree of sensitivity, exposure and adaptive capacity involving intrinsic and extrinsic factors (Nillson and Grelsson 1995).

Here, we developed an ecological vulnerability assessment tool for PAs based on Big Earth Data. The approach was applied to case studies of IUCN PAs in the basins of the Aral Sea, the Caspian Sea, the Black Sea and the Lake Baikal (Three Seas and One Lake, TSOL) (Figure 1). The TSOL region has attracted the attention of the international community, because of a variety of crisis from ecological environment or resource development. For example, the severe shrinkage of the Aral Sea and water resources coordination issues in the upstream and downstream regions lead to the serious ecological crisis in the basin of Aral Sea (Wu et al. 2020). As the largest lake in the world, the Caspian Sea is located at the junction of Eurasia and has a large number of oil and gas resources transformed from marine sediments (Konyushkova et al. 2021; Pamucar et al. 2021). The Black Sea is a large and anoxic marine system, which is faced with the problem of huge environmental pollution, and the Black Sea Strait is also an important navigation route connecting Europe, Asia, the Black Sea and the Mediterranean Sea (Karsli 2021; Pojar et al. 2021). The reserve and quality of fresh water resources of the Lake Baikal is second to none in the world, and its basin is facing huge ecological risks brought by the development of mineral and forest resources (Brown et al. 2021; Swann et al. 2020).

Figure 1. The normalized difference vegetation index (NDVI) of the study area and the distribution of protected areas.

2. Study area and datasets

2.1. Study area

The study area spans multiple climatic zones including the tropics, subtropics, warm temperate, and cold temperate zones (Figure 1). The study area is of international concern because it covers a large area, with multiple land cover types and a wide disparity in national economic levels and cultures. For example, the Aral Sea used to be the fourth largest lake in the world (Bekchanov, Ringler, and Bhaduri 2018), but uncontrolled water intake led to a sharp reduction of more than 90% of the lake area, and soil salinization led to ecosystem damage.

2.2. Datastes

The remote sensing data used in this study were obtained from MOD13A3 and MCD12Q1 (resolution of 1 km and 500 m, respectively), Operational Linescan System nighttime light images (1 km resolution), and the products of ASMR2/ASMRE Soil Moisture (25 km resolution). The NASA Shuttle Radar Topographic Mission (SRTM v4.1, 90 m resolution) provided digital elevation data (DEMs). The precipitation data were from the CRU TS v. 4.01 dataset at 0.5° resolution, which was released on 20 September 2017. The data details are listed in Table 1.

Table 1. Basic information of the products in the paper.

Products	Year	Resolution	Data sources
CRU TS v. 4.01	2017	0.5°	https://crudata.uea.ac.uk/cru/data/hrg/
SRTM v4.1	2008	90 m	http://srtm.csi.cgiar.org/
Night Light Index Products	2000–2015	1 km	https://www.ngdc.noaa.gov/eog/dmsp/
			https://www.ngdc.noaa.gov/eog/viirs/
MOD13A3		1 km	https://ladsweb.modaps.eosdis.nasa.gov/
MCD12Q1		500 m
ASMR2/ASMRE	2012/07/02	25 km	http://nsidc.org/data/amsre/index.html
WDPA	2020	-	https://www.protectedplanet.net/
KBAs, EBAs	2020	-	http://www.keybiodiversityareas.org/
IFLs	2013	-	http://www.intactforests.org/
G200	2002	-	https://www.worldwildlife.org/publications/global-200
GLWD	2004	1 km	http://www.wwfus.org/science/data.cfm

3. Methods

3.1. Calculation of the ecological vulnerability

Many scholars have studied ecological vulnerability from the perspectives of exposure, sensitivity and adaptability (Kling et al. 2020; Mumby et al. 2014; Nguyen et al. 2016; White 1974). This study used the advantages of Big Earth Data (Guo 2018; Guo et al. 2020; Wang et al. 2019), developed and enriched the meaning, contents and methods of calculating exposure, sensitivity and adaptability, and quantitatively expressed the spatial–temporal distribution characteristics and patterns of ecological vulnerability. Based on the analytic hierarchy process (AHP) and experts mark, this paper established the index evaluation system according to the specific situation of ecological vulnerability (Table 2).

Table 2. The evaluation index system for ecological vulnerability.

System layer (W1)	Structure layer (W2)	Datum layer (W3)	Acronyms	W4
Exposure (0.4)	Rainfall (−0.5)	Precipitation Coefficient	PC	0.200
	Topography (+0.2)	Topographic Elevation Difference Index (0.5)	TED	0.040
		Slope Difference Index (0.5)	SD	0.040
	Human interference (+0.3)	Human Interference Intensity Index	HII	0.120
Sensibility (0.3)	Soil moisture (−0.5)	Soil Moisture Index	SM	0.150
	Ecosystem vitality (−0.5)	Ecosystem Vitality Index	EV	0.150
Adaptability (0.3)	Protected area Index (−0.5)	Area Coefficient of Protected Area (0.5)	ACPA	0.075
		Number Coefficient of Protected Area (0.5)	NCPA	0.075
	Degree of national coordination (−0.5)	National Development Index	ND	0.150

Combining remote sensing (RS) data and geographic information systems (GIS) technology, this paper assessed the EVI (Ecological Vulnerability Index) from 2001 to 2015 and explained the variations of the EVI by various data. The details of the EVI system are listed in Appendix 1.

3.2. Standardized calculation and multi-criteria evaluation

The values of each indicator were standardized to reflect the positive/negative contribution of each variable towards the objective function. In the case of positive contributions to ecological vulnerability, the original X_i values were converted into a range of F_(max) values from 0 to 1, according to the following formula (Wu et al. 2020): (1) $F_{\left(max\right)}=\left(X_i-X_{min}\right)/\left(X_{max}-X_{min}\right)$(1) where F_(max) are the normalized data, X_i are the data to be normalized, X_min is the minimum value assumed by the variables and X_max is the maximum value.

In contrast, a negative contribution to ecological vulnerability was calculated using the following formula, where the variables have the same meaning as Equation (1). (2) $F_{\left(min\right)}=\left(X_{max}-X_i\right)/\left(X_{max}-X_{min}\right)$(2)

Finally, the EVI was given by the sum of the standardized values as described by the following formula: (3) $EV{I}_i=\sum _{i=0}^{n}XjWj$(3) where n is the number of variables, and W_j is the weight associated to the j_th elementary criterion with i=1 and 0≤W_j≤1.

3.3. Classification of the degree of ecological vulnerability

Synthesis maps of DEV were produced and classified as serious, high, medium, low or normal based on the natural interval classification method (Nguyen et al. 2016). The details of the classification of DEV were shown in Table 3.

Table 3. The classification and grading of degree of ecological vulnerability compared with the ecological vulnerability index.

Degree of ecological vulnerability	Normal	Low vulnerability	Medium vulnerability	High vulnerability	Serious vulnerability
Ecological vulnerability index	(-∞, −0.55]	(−0.55),−0.45]	(−0.45, −0.35]	(−0.35, −0.25]	(−0.25,+∞]
Description	Healthy ecosystem	Ecosystems are with low volatility	Ecosystems are at an acceptable volatility level	Ecosystems are difficult to sustain	Ecosystems are at the risk of collapse

4. Results

4.1. Distribution of PAs within the TSOL region

There were 5, 296 PAs in the TSOL region, with a total area of 5.47×10⁵ km², representing 10.28% of the total land area of the basins (Table 4). The PAs in the Lake Baikal basin (23.55%) covered the largest area, followed by the Black Sea basin (8.24%). The PAs in the Aral Sea basin and the Caspian Sea basin covered relatively less area, at 5.57% and 3.74%, respectively. In terms of the numbers of PAs, the Black Sea basin had the highest percentage, accounting for 95.11% of the total number. However, the PAs in the Black Sea basin were relatively small with an average size of 39.94 km². The PAs in Lake Baikal basin were the largest on average (4370.97 km²), followed by the Aral Sea basin (1089.02 km²) and then the Caspian Sea basin (828.77 km²).

Table 4. Numbers and area of protected areas (PAs) in the TSOL region.

Basin	Area of basin (×10^4 km^2)	Area of PAs (×10^4 km^2)	Number of PAs	Area Ratio (%)
The Aral Sea	160.37	8.93	82	5.57
The Black Sea	244.29	20.12	5037	8.24
The Caspian Sea	323.44	12.1	146	3.74
The Lake Baikal	57.52	13.55	31	23.55
Sum	785.62	54.7	5296	10.28

4.2. Biodiversity important places (BIPs) within the TSOL region

Figure 2 shows the distribution of BIPs, which were created by combining KBAs, endemic bird areas (EBAs), intact forest landscapes (IFLs), global 200 ecoregions (G200), and the global lakes and wetlands database (GLWD) (Davis et al. 1997; Eken et al. 2004; Olson and Dinerstein 2002; Stattersfield et al. 1998; Potapov et al. 2008; Zheng et al. 2015). The BIPs within the TSOL region covered an area of 3.61 million km², accounting for 51.05% of the total area.

Figure 2. The distribution of biodiversity important places (BIPs) inside and outside protected areas by overlaying key biodiversity areas (KBAs), endemic bird areas (EBAs), intact forest landscapes (IFLs), global 200 ecoregions (G200) and the global lakes and wetlands database (GLWD).

In these basins, the BIPs covered the greatest area in the Aral Sea basin (78.63%), followed by the Lake Baikal basin (53.39%), the Black Sea basin (37.28%) and the Caspian Sea basin (34.91%). Of these BIPs, 15.79% were within PAs with a total area of 3.73×10⁵ km². The Lake Baikal basin had the highest coverage (38.65%), followed by the Black Sea basin (12.73%), the Caspian Sea basin (7.51%) and the Aral Sea basin (4.26%) (Table 5).

Table 5. The area and ratio of biodiversity important places (BIPs) in the different basins.

Basin	Area of BIPs (×10^4 km^2)	The area ratio between BIPs and its basin	Area of BIPs in PAs (×10^4 km^2)	The proportion of BIPs covered by PAs
The Aral Sea	126.09	78.63%	5.37	4.26%
The Black Sea	91.07	37.28%	11.59	12.73%
The Caspian Sea	112.91	34.91%	8.48	7.51%
The Lake Baikal	30.71	53.39%	11.87	38.65%
Sum	360.78	51.05%*	37.31	15.79%*

4.3. DEV within the TSOL region

The DEV of the TSOL region ranged from −0.13 to −0.69 and showed spatial stability with some minor volatility during 2001 and 2015 (Figure 3). The DEV of the Aral Sea basin was relatively low in the upper reaches and increased in the middle and lower reaches of the river.

Figure 3. Degree of ecological vulnerability within the TSOL region from 2001 to 2015.

For the Lake Baikal basin, the most ecologically vulnerable areas were mainly in the south, while regions with lower levels of vulnerability were concentrated in the east and north. The Black Sea basin and the Caspian Sea basin both showed a gradual increase in vulnerability from north to south and from west to east. The ecological condition in the sandy plain areas of the southern Black Sea basin and the flatlands in the eastern Caspian Sea basin was highly vulnerable. The vulnerability fluctuations in the Aral Sea basin were mainly concentrated in the central and western regions, while the vulnerability fluctuations in the Aral Sea basin were mainly concentrated in the southern part of the region.

4.4. DEV inside and outside the BIPs

In 2015, the DEV of the BIPs was generally at a medium level, with the highest vulnerability in the Aral Sea basin, and lowest in the Black Sea and Lake Baikal basins. There was a high DEV in the PAs in the Caspian Sea basin, and it was higher than that outside the PAs. In contrast, the DEV in the Aral Sea basin was high outside the BIPs. The high vulnerability areas require special attention to enhance ecological restoration (Table 6). From 2001 to 2015, the DEV of the BIPs remained steady, and usually DEV outside the BIPs was higher than that inside the BIPs. In 2015, DEV inside and outside the BIPs began to level off. In the Black Sea basin, the DEV of the BIPs showed a trend of improvement although there were periodic fluctuations. In the Aral Sea basin, the DEV outside BIPs remained at a high level for many years, with a slight improvement in 2015.

Table 6. DEV (degree of ecological vulnerability) of BIPs (biodiversity important areas) inside and outside protected areas (PAs) among the different basins.

Year	Basin	DEV of BIPs in PAs	DEV of BIPs outside PAs	Total
2015	Aral Sea	Medium	High	High
	Black Sea	Low	Medium	Low
	Caspian Sea	High	Medium	Medium
	Lake Baikal	Low	Low	Low
	Average	Medium	Medium	Medium
2010	Aral Sea	High	Serious	High
	Black Sea	Medium	High	Medium
	Caspian Sea	High	High	Medium
	Lake Baikal	Low	Medium	Low
	Average	Medium	High	Medium
2005	Aral Sea	High	Serious	High
	Black Sea	Low	High	Low
	Caspian Sea	High	High	Medium
	Lake Baikal	Low	Medium	Low
	Average	Medium	High	Medium
2001	Aral Sea	High	Serious	High
	Black Sea	Medium	High	Medium
	Caspian Sea	High	High	Medium
	Lake Baikal	Low	Medium	Low
	Average	Medium	High	Medium

4.5. Spatiotemporal patterns of DEV in PAs

Overall, the DEV in PAs generally was stable at a medium level between 2001 and 2015, particularly for the major PA types such as national parks and habitat/species management areas. The DEV in a few types of PAs fluctuated (Figure 4).

Figure 4. Maps of the spatiotemporal variation patterns of DEV (degree of ecological vulnerability) in PAs (protected areas) within the TSOL region from 2001 to 2015.

The Caspian Sea basin remained at a moderately high level and the DEV of protected landscape/seascape was medium to high. In contrast, the Black Sea basin had a moderately low DEV level, as did wilderness areas. The ecological situation as indicated by the DEV in the following categories of PAs improved: the Lake Baikal basin (I_a), the Caspian basin (V), and the Black Sea basin (III, IV, and VI). The PAs requiring additional attention included: (1) high DEV in the Caspian basin (I_a and V) and the Aral Sea basin (V); (2) medium and high DEV in the Aral Sea basin (I_a) and the Caspian Sea basin (II, III, IV and unclassified); and (3) the aggravated DEV in the Black Sea basin (I_b, V), the Aral Sea basin and the Lake Baikal basin (Unclassified) (Table 7 and Appendix 2).

Table 7. Degree of ecological vulnerability in the studied protected areas (grouped by category) from 2001 to 2015. I_a: strict nature reserve, I_b: wilderness area, II: national park, III: natural monument or feature, IV: habitat/species management area, V: protected landscape/seascape, VI: protected area with sustainable use of natural resources.

Year	Basin	Ia	Ib	II	III	IV	V	VI	Unclassified	Average
2015	The Aral Sea	High	-	Medium	-	Medium	Serious	-	Medium	Medium
	The Black Sea	Low	Normal	Low	Low	Low	Low	Low	Low	Low
	The Caspian Sea	Serious	-	High	High	High	High	-	High	High
	The Lake Baikal	Normal	Low	Medium	Low	-	-	Medium	Low	Low
	Average	Medium	Low	Medium	Medium	Medium	High	Medium	Medium	Medium
2010	The Aral Sea	High	-	Medium	-	Medium	Serious	-	High	Medium
	The Black Sea	Low	Low	Low	Medium	Medium	Medium	Medium	Low	Low
	The Caspian Sea	Serious	-	High	High	High	High	-	High	High
	The Lake Baikal	Low	Low	Medium	Low	-	-	Medium	Medium	Medium
	Average	Medium	Low	Medium	Medium	Medium	High	Medium	Medium	Medium
2005	The Aral Sea	High	-	Medium	-	Medium	Serious	-	Medium	Medium
	The Black Sea	Low	Normal	Low	Low	Medium	Low	Low	Low	Low
	The Caspian Sea	Serious	-	High	High	High	High	-	High	High
	The Lake Baikal	Low	Low	Medium	Low	-	-	Medium	Low	Medium
	Average	Medium	Low	Medium	Medium	Medium	High	Medium	Medium	Medium
2001	The Aral Sea	High	-	Medium	-	Medium	Serious	-	Medium	Medium
	The Black Sea	Low	Normal	Low	Medium	Medium	Low	Medium	Low	Low
	The Caspian Sea	Serious	-	High	High	High	Serious	-	High	High
	The Lake Baikal	Low	Low	Medium	Low	-	-	Medium	Low	Medium
	Average	Medium	Low	Medium	Medium	Medium	High	Medium	Medium	Medium

5. Discussion

According to our former research results (Wu et al. 2020), the ecological carrying capacity (ECC) of the Aral Sea basin is getting better during 2001–2018, and the ECC of the upstream region is better than that of the downstream region. In this paper, the EVI of the Aral Sea basin is getting better during 2001–2015, and the spatiotemporal distribution law of EVI is similar to that of ECC. A water trading system can improve inflows to the Aral Sea but would require significant compensation for agricultural producers (Bekchanov, Ringler, and Bhaduri 2018). In the last three decades, several ecological factors were changed and affected the most important commercial fish stocks in the Caspian Sea, and the stock of the stellate sturgeon (Acipenser stellatus Pallas) was being overfished for the whole years and critically endangered (Fazli et al. 2021). However, the vulnerability of terrestrial protected areas in the Caspian Sea basin has improved, it was shown as that the DEV of BIPs outside terrestrial PAs has changed from high to medium in this paper.

The EVI of the Black Sea basin showed a gradual increase in vulnerability from north to south and from west to east, and the ecological condition in the sandy plain areas of the southern Black Sea basin was highly vulnerable. The result is consistent with a previous paper, which shows that the Black Sea coastline of Georgia is evaluated as the region, characterized with the highest grade of ecological vulnerability (Lagidze et al. 2017). In the last century, the southern Europe have been impacted by a combination of anthropogenic and climatic pressures, which have rapidly evolved to change the ecological status of freshwater and coastal systems. The southern Black Sea basin is differently impacted by anthropogenic and climatic forces and this will influence their vulnerability to future changes of flow regime and water quality (Micheli et al. 2013). As lakes warm worldwide (Meyer et al. 2017; Swann et al. 2018), climate change and human Interference may alter not only plankton community structure and abundance of the Lake Baikal, but also the EVI of the total ecosystem in the basin. Compared with former approaches (Konyushkova et al. 2021), this paper integrated the information about forests, ecological zones and wetlands. This study focused on assessing the condition and trends of PAs from the perspective of ecosystem vulnerability, and provided some support for the optimization of spatial pattern of the PA network. When PA network is discussed, the regional sustainable development should also be considered (Anderson et al. 2017; Diaz et al. 2020; Thiault et al. 2020).

6. Conclusion

The approach developed in this paper illustrates how to derive a more integrated understanding of the condition of linked social-ecological systems for ecological vulnerability analysis on a large scale. Strengthening international cooperation is an important means to break the bottleneck of ecological environment and resource endowment in these important international basins. Big Earth Data science shows great vitality in protected area ecological vulnerability assessment and regional sustainable development. The spatial pattern of ecological vulnerability in the TSOL region was relatively stable during 2001–2015. The Caspian Sea basin and protected landscape/seascape (category V) showed high ecological vulnerability, while the Black Sea basin and wilderness areas (category I_b) showed low ecological vulnerability. The effective measure to conserve and maintain biodiversity is to optimize the quality of PAs on the basis of ensuring the coverage of protected areas. Big Earth Data could be used to identify the priority areas for ecological restoration by quickly and accurately identifying areas of ecological vulnerability over a large scale in future.

Data and materials availability

All data needed to replicate the conclusions in the paper are presented in the paper and/or the Supplementary Materials, or the references cited. Additional data related to this paper may be requested from the authors.

Acknowledgements

We thank the anonymous reviewers for this manuscript. We also thank Leonie Seabrook, PhD, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences (grant number XDA19030104, XDA19090120 and XDA19090124).

Appendices

Appendix 1. The indicators and the detailed information used to evaluate ecological vulnerability.

Table

Dimension	Indicator	Equation	Description
Exposure	Precipitation Coefficient	$P{C}_i=\sum _{j=1}^{n}PCj.S0$	PCi is the precipitation coefficient of sub-basin i, and PCj is the annual precipitation of grid j in sub-basin i, and S0 is the area of grid.
	Topographic Elevation Difference Index	$TE{D}_i={\displaystyle \frac{\sum _{j=1}^{n}TEDj}{n.TEDmax.}.fhi}$, $f_{hi}={\displaystyle \frac{{\sigma }_{hi}}{{\mu }_{hi}}}$	TEDi is the topographic elevation difference index of sub-basin i, TEDj is the average elevation of grid j in sub-basin i, n is the number of sub-basin, TEDmax is the maximum elevation of sub-basin i. Where fhi is the topographic elevation difference coefficient of grid in sub-basin i, σhi is the topographic elevation variance of grid in sub-basin i,μhi is the average elevation of grid in sub-basin i.
	Slope Difference Index	$S{D}_{i=}{\displaystyle \frac{\sum _{j=1}^{n}SDj}{n.Smax}{f}_{lsi}}$, $f_{ls}i={\displaystyle \frac{{\sigma }_{lsi}}{{\mu }_{lsi}}}$	SDi is the slope difference index of sub-basin i, SDj is the average slope of grid j in sub-basin i. Where flsi is the slope difference coefficient of grid in sub-basin i, σlsi is the slope variance of grid in sub-basin i, μlsi is the average slope of grid in sub-basin i.
	Human Interference Intensity Index	$HI{I}_i={\displaystyle \frac{\sum _{k=1}^m\sum _{j=1}^{n}Wk.HIIj}{Si}}$	HIIi is the human interference intensity index of sub-basin i, HIIj is the area of land use/cover of human interference in grid j of sub-basin i, Wk is the weight of land use/cover k of human interference, Si is the area of sub-basin i.
Sensibility	Soil Moisture Index	$S{M}_i={\displaystyle \frac{\sum _{j=1}^{n}SMj.Sj}{Si}}$	SMi is the soil moisture index of sub-basin i, SMj is the average value of soil moisture index in grid j of sub-basin i, Sj is the area of grid j, Si is the area of sub-basin i.
	Ecosystem Vitality Index	$E{V}_i={\displaystyle \frac{\sum _{j=1}^{n}NPPj.Sj}{NPPmax.Si}}$	EVi is the ecosystem vitality index, NPPj is the average value of net primary productivity in grid j of sub-basin i, Nppmax is the maximum value of net primary productivity in sub-basin i,Sj is the area of gird j, and Si is the area of sub-basin i.
Adaptability	Protected Area Index	$P{A}_i={\displaystyle \frac{\sum _{j=1}^{n}PASj}{Si}+\frac{PAPi}{PAPmax}}$	PAi is the protected area index of sub-basin i, PASj is the area of protected areas based on polygon in grid j of sub-basin i, Si is the area of sub-basin i,PAPi is the number of protected areas based on point in sub-basin i, PAPmax is the max number of protected areas in sub-basin i.
	National Development Index	$N{D}_i={\displaystyle \frac{\sum _{j=1}^n(Li-Lmin).S0}{(Lmax-Lmin).Si}}$	NDi is the national development index of sub-basin i, Lmax and Lmin are the maximum and the minimum light index in sub-basin i, respectively. S0 and Si are the area of grid j and sub-basin i.

Appendix 2.The proportion (%) of different categories of protected areas in the study area. Ia: strict nature reserve, Ib: wilderness area, II: national park, III: natural monument or feature, IV: habitat/species management area, V: protected landscape/seascape, VI: protected area with sustainable use of natural resources.

Table

Basin	Ia	Ib	II	III	IV	V	VI	Unclassified
The Aral Sea	10.27	0.00	34.47	0.00	12.52	0.23	0.00	42.51
The Black Sea	2.70	0.04	3.38	0.24	5.60	2.90	0.07	85.06
The Caspian Sea	7.18	0.00	10.64	0.03	14.04	20.89	0.00	47.23
The Lake Baikal	4.79	11.72	22.83	0.36	0.00	0.00	0.86	59.43
Average	6.24	2.94	17.83	0.16	8.04	6.01	0.23	58.56