Ecosystem vulnerability to extreme climate in coastal areas of China

Climate change has far-reaching impacts on ecosystems and the frequency and intensity of extreme global climate events have been increasing over the past century; therefore, assessing ecosystem vulnerability to extreme climate change is critical for sustainable and adaptive ecosystem management. As a climatically sensitive region, coastal China is currently experiencing significant environmental changes. To identify how extreme climate affects ecosystem vulnerability, we calculated and analyzed the spatiotemporal variation in extreme climates, net primary productivity (NPP), and spatial characteristics of ecosystem vulnerability to extreme climate change, and discussed the response characteristics of different ecosystems to extreme climate events based on meteorological data and NPP (1986–2015). The results demonstrated that (1) coastal China has become increasingly warmer over the last thirty decades but the precipitation trend is different in the north and south: precipitation increased in the south and decreased in the north. (2) NPP is rising overall, with the forest ecosystem growing the fastest, particularly since 2010. (3) The ecosystem vulnerability of coastal areas in China is mainly classified as mild or non-vulnerable. However, there were apparent differences in the vulnerability of different ecosystems, with dry land and shrub ecosystems having the highest mean vulnerability. (4) The effects of extreme climates on the vulnerability of different ecosystems and ecosystems in different habitats vary. Overall, rising extreme temperatures can significantly increase the ecosystem vulnerability in the coastal areas of China. The paddy field ecosystem was more influenced by extreme temperatures than other ecosystems, with the southern paddy field ecosystem more influenced than the northern paddy field ecosystem. Our study advances the understanding of vegetation dynamics and their driving mechanisms and provides support for scientifically informed ecological management practices in coastal China.


Introduction
The structure, composition, and function of terrestrial ecosystems are influenced by climate change, particularly extreme climatic events (Phillips et al 2009, Doughty et al 2015, Fu et al 2018).Therefore, climate change poses enormous challenges to the protection of terrestrial ecosystems and practical protective measures must be taken to meet these challenges (McClanahan et al 2008, Dawson et al 2011, Lemieux and Scott 2011, Anderson et al 2014).The increasing frequency and intensity of extreme climate events expose vulnerable areas of terrestrial ecosystems to greater risk (Araújo et al 2011).Different ecosystems have unique risks (Yongxiang et al 2015).Therefore, accurately understanding and assessing the vulnerability of diverse ecosystems to extreme climates are critical for mitigating the threat of climate change and realizing sustainable and adaptive ecosystem management (Tse-ring et al 2010, Hughes 2011).
Ecosystem vulnerability refers to the vulnerability of an ecosystem to damage, stress, and lack of adaptation associated with environmental and social change (Alwang et al 2001, Adger 2006, Butler et al 2016).
Ecosystem vulnerability is affected by natural factors, such as geographical location (Li et al 2018), and by climate extremes (Cardona et al 2012, Handmer et al 2012, Olds et al 2014).Extreme climates are defined as statistically rare or unusual climatic periods, in which the function of an ecosystem is altered well outside the bounds of its threshold (Reichstein et al 2013).For example, high temperatures, droughts, and fires are extreme climates associated with abnormal temperatures and precipitation.Changes in temperature and precipitation have resulted in a decrease in forests (Choat et al 2012).During drought, the water transport system produces a large number of gas emboli, which reduces the potential water supply capacity (Nourtier et al 2014), eventually, causing trees to die (Choat et al 2012).Du et al (2019) found that spring vegetation phenology changes significantly in arid areas because of the interactions between temperature and precipitation under extreme climates.Ultimately, the ecosystem functions in the area have changed.Previous studies have focused on the impacts of climate change on ecosystem vulnerability (Lim et al 2006, Ravindranath et al 2006, Xiao-Ying et al 2013, Meng et al 2016) and climate change is considered a threat to vegetation that affects ecosystem vulnerability.For example, Lindner et al (2010) found that global warming and an increase in carbon dioxide have a positive impact on forests while increasing the risk of drought and fire.Moreover, there were more adverse than positive effects.Some studies have found that frequent and persistent droughts due to climate change cause diseases in the forest canopy, change the function of crown leaf structures (Lloret et al 2004, Cano et al 2013, Matusick et al 2013, Saatchi et al 2013), and increase the mortality rate of trees (Rouault et al 2006, Klos et al 2009, Gustafson and Sturtevant 2013, Young et al 2017, Choat et al 2018).Nandintsetseg et al (2021) conducted risk and vulnerability assessments on Inner Mongolian grasslands for 40 years  based on probability theory and found that the probability of hazardous drought and ecosystem vulnerability increased with more frequent droughts that are dangerous to climate change.Chen et al (2014) comprehensively evaluated the vulnerability of grassland ecosystems on the Qinghai-Tibet Plateau and found that the vulnerability of the marginal area of the Qinghai-Tibet Plateau was higher than that of the regional area and that different grassland types had different responses to climate change.
Most studies on ecosystem vulnerability to climate change focused on inland areas, while few scholars have studied the ecosystem vulnerability of China's coastal areas.In addition, most studies focused on single ecosystems.In contrast, the present study focuses on the vulnerabilities of different ecosystems to extreme climate events and aims to reveal the regional characteristics of the impact of extreme climate events on ecosystem vulnerability by studying the relationship between extreme climate indices and ecosystem vulnerability.This study provides a reference and basis for the adaptation to and mitigation of climate change; ecological environment protection in coastal areas; promotion of economic and social health; and sustainable development of the environment in coastal areas.

Research area
China's coastal areas are located between the western Pacific Ocean and Eurasia, between 104 • 26 ′ E and 125 • 47 ′ E and 3 • 20 ′ N and 43 • 29 ′ N. The land area of China's coastal areas covers approximately 1.3 million km 2 , accounting for approximately 14% of the national land area.The natural conditions of China's coastal regions are complex, consisting of marine and land transition areas (figure 1(a)).This area belongs to the East Asian monsoon region.The north-tosouth span of the region is considerable, the span of north-south latitude is about 40 • , with a distance of 5500 km; therefore, the north and south regions are quite different and vary owing to the different heat transfers.The vegetation types in this region also vary spatially.Northeast, Jiangnan, and South China are dominated by forest ecosystems, whereas northern, Huanghuai, and Jianghuai regions are dominated by farmland ecosystems (figure 1(b)).
China's coastal areas include 13 administrative regions (excluding Taiwan and the South China Sea Islands).The precise geographical locations and meteorological stations are shown in figure 1(a).

Data
The 'monthly 1 km raster data set of net primary productivity (NPP) of terrestrial ecosystems north of 18 • N (1985-2015)' was downloaded from the Global Change Science Research Data Publishing System (Chen 2019) and the annual NPP data were calculated from the monthly data.NPP data for a long time series of terrestrial ecosystems in China were calculated using the CASA model based on meteorological, soil, land cover, and vegetation index data from 1986 to 2015.
Meteorological data were obtained from the National Meteorological Information Centre of the China Meteorological Administration.Considering the continuity of the data and historical record data of more than 50 years as the standard, the RClimdex model was used for missing data processing and quality control.Finally, meteorological data from 175 meteorological stations were selected for extraction and calculation of the extreme climate index.
The spatial distribution of the ecosystem adopted the '2015 China Land Use and Cover Dataset (CLUDS) 1 km' produced by the Chinese Academy of Sciences, superimposed it with the vegetation map of China, made a comprehensive analysis with the climate elements, and obtained the spatial distribution of the ecosystem in the coastal areas of China.

Extreme climates indicators
The extreme climate index is a statistical method used to describe and quantify the probability, intensity, and duration of extreme weather events, based on historical observation data (www.eca.knmi.nl).At present, threshold is the most commonly method in climate extreme change research internationally.We arranged the climate elements in ascending order, obtained 10th and 90th percentiles value as the percentage threshold.From the extreme climate indices recommended by the ETCCIDI, 11 were selected for this study (table 1).This study used the RClimDex model for the exponential calculations (Karl et al 1999, Peterson et al 2001).

Ecosystem vulnerability
The selection of indicators is vital for the evaluation to reveal the internal changes in the system and its response characteristics to external environmental conditions, rather than simply describing the status quo (Li et al 2005(Li et al , 2012)), and to avoid the uncertainties introduced by the multi-indicator system.In this study, NPP was selected as the ecosystem function index to evaluate the characteristics of ecosystem vulnerability.According to the relevant definition proposed by the Intergovernmental Panel on Climate Change, vulnerability is a function of the (1) This study considered the annual fluctuation in NPP and its direction as an indicator of the sensitivity and adaptability of the ecosystem respectively (Li et al 2012).The formula is as follows: where F i is the value of NPP in year i and F is the average NPP over n years.The S value can reflect the degree of discrete of the annual scale NPP; the larger the S value, the stronger the inter-annual volatility of the NPP, the greater the sensitivity to climate change.
On the contrary, the smaller the S value, it means that NPP is not sensitive to climate change.This paper regards the changing direction of the annual fluctuation of NPP as the adaptability of the ecosystem (Li et al 2012).The formula to quantify adaptability (A) is as follows: where x i and x j are the NPP values at times i and j (j > i), respectively, n is the length of the data sample, and t i is the number of data points in group i.In equation ( 3), if A is greater than 0, it indicates that the time sequence of NPP shows an upward trend, indicating that the NPP is gradually increasing to adapt to climate change.If A is less than 0, it indicates that the time sequence of NPP shows a downward trend, indicating that the adaptability of NPP to climate change is poor and gradually decreases.
The method to classify ecosystem vulnerability levels is the natural breaks classification method from ArcGIS.Based on natural groupings inherent in the data, the method is a very effective data analysis and visualization technique.It uses the characteristics of the data itself, based on the continuous value space, the data is divided into multiple independent levels, and the differences between the levels are the most obvious.The advantage of natural breaks classification method is that it can effectively avoid the flattening effect caused by the fixed breakpoint method, and has strong adaptability, so that the data aggregation of all levels in the visualization is as uniform as possible, so as to better express the real distribution characteristics.Because natural breaks classification places clustered values in the same class, this method is good for mapping data values that are not evenly distributed (De Smith et al 2007).

Variation characteristics of extreme climates events
The changes in extreme climate indices in coastal areas showed a consistent trend and the extreme temperature indices (figure 3), TXx, TNx, TXn, and TNn, which characterize the daily temperature extremes, all showed an increasing trend over time.The TNn and TNx indices showed significant increasing trends; however, only the TXx index in the north showed a partial downward trend.However, the number of warm days (Tn90p), warm nights (Tx90p), cold days (Tx10p), and cold night (Tn10p) exhibited different trends: Tx90p and Tn90p increased significantly, indicating that the number of high-temperature days in the study area significantly increased over time.Tx10p and Tn10p showed downward trends, with Tn10p decreasing significantly.The results indicated that the number of cold and cold nights in the study area decreased with time.DTR, representing the daily temperature range, decreased significantly with time, indicating that the temperature in the study area increased during the day and night, and the rate of  increase during the night was greater than that during the day.Compared with the extreme temperature index, the extreme precipitation indices, Rx1day and Rx5day, showed insignificant increasing trends and there was a significant difference between the south and north.The extreme precipitation index increased in the south, whereas it decreased in the north.
The trends of the most extreme temperature indices in the coastal areas of China exhibited a certain degree of co-directional persistence.Figure 2 shows that the index of characterized extreme climate events at high temperatures will continue to increase in the future.In contrast, the index of extreme climate events at low temperatures continued to show a downward trend, suggesting that coastal areas will continue to warm.The trend of the extreme precipitation index in the coastal regions of China had longterm autocorrelation characteristics but the degree of continuous change was generally weak.The results indicate that the coastal areas of North China, Huanghuai, and Jianghuai will become arid, whereas the northeastern coastal areas will become wet.

Trends in ecosystem NPP
As there are very few NPP weather sites to monitor shrub ecosystems, the NPP of shrub ecosystems was calculated in combination with the NPP of forest ecosystems.The results showed that the NPP of all four types of ecosystems showed increasing trends, of which the increasing trend of dry land was most apparent and the forest ecosystem showed a significant increasing trend after 2010 because of the progress of measures such as the conversion of cropland to forest and the greening of barren hills.

The spatial variation characteristics of the NPP
From 1986-2015, the NPP in China's coastal areas showed an increasing trend from northeast to southwest (figure 4(a)).The NPP significantly increased in western Liaoning Province, southern Hebei Province, Shandong Province, and the Pearl River Delta region (figure (c)) but considerably decreased in Zhejiang Province and Fujian Province.The NPP was higher in the forest ecosystem area in the south and lower in the dry land and grassland ecosystem areas in the north.

Spatial patterns of ecosystem vulnerability
Relatively speaking, the ecosystem vulnerability of coastal areas in China was mainly not vulnerable and mild vulnerability (figure 5), accounting for 30% and 27.5% of coastal areas, respectively, followed by 19% moderate vulnerability, 12.5% severe vulnerability, and 11% extreme vulnerability.The areas with relatively high vulnerabilities were in North China and Huanghuai.

Relationship between extreme climates and vulnerability of different ecosystems
The extreme climate indices TXn, TNn, Tn90p, Tx90p, and DTP and the vulnerability of the southern paddy field ecosystems were more correlated, with slopes of 0.23, 0.16, 0.28, 0.40, and −0.88, respectively, and passed the significance test, with a correlation level of 0.1.Compared with the vulnerability of the southern paddy field ecosystem, the extreme temperature indices TXx, TXn, and TNn showed significantly stronger positive correlations with the vulnerability of the northern paddy field ecosystem, with slopes of 0.59, 0.21, and 0.21, respectively (table 2).
The extreme temperature indices TNx, Tn10p, and Tn90p were significantly correlated with the vulnerability of the dryland ecosystem in the south, with slopes of 3.61, −0.80, and 0.46, respectively.Compared with the south, all extreme climate indices were weakly correlated with the dryland ecosystems in the north, with significance levels not reaching 0.1.
The extreme air temperature indices TXn, TNn, and Tx90p were significantly and positively correlated with the vulnerability of the forest ecosystem in the south, with slopes of 0.26, 0.22, and 0.41, respectively.Compared with the forest ecosystem in the south, the correlation between the extreme climate index and vulnerability of the forest ecosystem in the north was weak and not significant.
Due to data limitations, there was no northsouth analysis of shrub and grassland ecosystems.The The absolute value of the correlation coefficient between the extreme climates index and grassland ecosystem vulnerability was concentrated in the range of 0.2-0.3 and the overall correlation was weak.Only Tx90p showed a strong correlation, reaching a significance level of 0.01.

Spatial distribution of ecosystem vulnerability
More than half of the ecosystem vulnerabilities in coastal areas were mild.The vulnerability of the ecosystems in northern and eastern Hebei, eastern Liaoning, and most areas south of the Jianghuai River in China was relatively low.The relatively low vulnerability of northern Hebei may be related to the implementation of large-scale vegetation protection and restoration projects, such as the control of sandstorm sources in the Beijing-Tianjin-Hebei region.
The northern part of Hebei Province belongs to the Yanshan-Taihang Mountain Ecological Zone and the eastern part belongs to the Coastal Ecological Protection Zone.The vegetation cover of the two regions during this period showed a continuously increasing trend and the surface vegetation cover increased (Zhang et al 2008, Wang et al 2018, Xu et al 2020b).Therefore, the ecosystem vulnerability in the region was relatively low.The relatively low ecosystem vulnerability in the eastern part of Liaoning Province was mainly due to the abundance of forest resources in the region, abundant hydrothermal energy in the Changbai Mountains, and relatively high NPP (Xu et al 2020c).The coastal areas of China south of Jianghuai (including) are mainly affected by the slow accumulation of daytime biomass in the north.In contrast, the rates of biomass consumption in the north-south coastal areas were not significantly different, and the biomass in the south was much higher than that in the north (Xu et al 2020a).The results indicated that the ecosystem vulnerability of southern coastal areas is lower than that of northern areas.Ecosystem vulnerability was high in western and southern Hebei and

Effects of extreme climates on different ecosystem types
Significantly correlated extreme climate indices differed for different ecosystems and differed with latitude for the same ecosystem type.The vulnerability of the paddy field ecosystem in the south was significantly correlated with TXn, TNn, Tn90p, Tx90p, and DTP.Rising extreme temperatures can lead to increased vulnerability of paddy field ecosystems and the average temperature in the south is higher than that in the north.When extreme temperatures increase, the respiration of vegetation in the paddy field ecosystem is strengthened and the loss of water increases, which is not conducive to vegetation production and ultimately leads to increased vulnerability.The vulnerability of northern paddy field ecosystems was significantly correlated with the extreme temperature indices TXx, TXn, and TNn because an increase in the frequency of temperature extremes accelerated water transportation in the northern region, where water was insufficient.The vegetation of the paddy field ecosystem was limited by water during the growth process; therefore, the vulnerability of the northern paddy field ecosystem increased.
The vulnerability of dryland ecosystems in the south was significantly correlated with the extreme temperature index TNx, Tn10p, and Tn90p.Of these indices, Tn10P, which characterizes extreme cold temperatures at night, had the strongest correlation with the vulnerability of dryland ecosystems in the south.When the temperature rises at night, the respiratory action of the vegetation is greatly enhanced and more organic matter is consumed, which makes the NPP fluctuation of vegetation more significant, eventually leading to increased vulnerability.The dryland ecosystem in the north was not significantly correlated with any extreme climate index, which may be due to factors such as artificial irrigation in the northern dry ecosystem.
The vulnerability of forest ecosystems in the south was positively correlated with TXn, TNn, and Tx90P.This may be explained by the fact that the increase in extreme temperature greatly enhances the respiration of vegetation, thus increasing the consumption of organic matter, causing the NPP to fluctuate and ultimately increasing vulnerability.Compared with forest ecosystems in the south, the correlation between the vulnerability of forest ecosystems in the north and the extreme climate index was low and not significant.
Owing to data limitations, shrub and grassland ecosystems were not analyzed separately from north to south but an overall analysis was conducted.The positive correlation between shrub ecosystem vulnerability and Tn10P indicates that an increase in Tn10p can have an obvious adverse effect on the production of the shrub ecosystem and eventually increase its vulnerability.There was a strong correlation between the vulnerability of the grassland ecosystem and Tx90p, which means that an increase in warm days will have a more significant adverse impact on the production of the grassland ecosystem with more water limitation and that the vulnerability of the grassland ecosystem will increase.Most studies on the effects of extreme climates on ecosystems have mainly characterized dynamic changes in ecosystems using the normalized difference vegetation index (NDVI), leaf area index, gross primary productivity, and NPP.However, this study used NPP to calculate the vulnerability index of ecosystems to characterize dynamic changes in ecosystems.The index used in this study is more comprehensive than those used in previous indices (e.g.NDVI) but the methodology for calculating ecosystem vulnerability needs to be improved.Population growth and economic development in China's coastal areas inevitably affect the natural ecosystems.This study used deseasonalization to remove as much interference as possible from human activities and other climate change overlays.However, these methods do not accurately reflect the actual dynamics of ecosystems.Future research should use a broader consensus approach to calculate ecosystem vulnerability and the pressure of human activities on ecosystem vulnerability and the proportion of natural and artificial factors affecting ecosystem vulnerability.

Conclusions
This study analyzed extreme climates, ecosystem function, ecosystem vulnerability, and the relationships between these three factors.The main conclusions are as follows: (1) Over the past 30 years, China's coastal areas have experienced extreme warming and the increase at night has been more significant than that in the daytime.
(2) From 1986 to 2015, the NPP in China's coastal areas showed an overall upward trend.The fastest growth rate was in forest ecosystems, which increased significantly since 2010.In the subregions, the average NPP in the south was higher than that in the north.However, the NPP in the Northeast China, North China, and Jiangnan Regions has increased in recent years owing to rapid urbanization, population growth, and rapid economic development, resulting in a downward trend in the NPP of their ecosystems.(3) The increasing trends of the extreme temperature index and extreme precipitation had positive effects on vegetation but there were some regional differences.(4) Extreme climates affect the vulnerability of the same ecosystem to diverse habitats or ecosystems.The increase in extreme temperatures significantly increased the vulnerability of coastal ecosystems.The increase in extreme precipitation increased the vulnerability of paddy field ecosystems but reduced the vulnerability of shrub and grassland ecosystems.The responses of forest and dryland ecosystems to extreme precipitation reflected the differences in habitat characteristics between the northern and southern regions.

Figure 1 .
Figure 1.Geographic location, meteorological station distribution, and land use distribution map of the study area.
system's ability to respond to climate change sensitivity and adaptation (McCarthy et al 2001, Field and Barros 2014), the more vulnerable systems are to climate change, the more sensitive ecosystems are, the less adaptable are.Therefore, vulnerability can be simply expressed as V = S − A.

Figure 2 .
Figure 2. Spatial distribution of extreme climates indices in coastal areas of China from 1986 to 2015.

Figure 3 .
Figure 3. 1986-2015 NPP trends in coastal areas and different ecosystem types in China.

Figure 4 .
Figure 4. NPP in coastal areas of China for many years and the trend.

Figure 5 .
Figure 5. Spatial pattern and histogram of ecosystem vulnerability in coastal areas of China 1986-2015.

Figure 6 .
Figure 6.Spatial distribution of vulnerability of different ecosystems.

Table 1 .
Precipitation and temperature indices.

Table 2 .
Correlation coefficient statistics between partial extreme climates indices and ecosystem vulnerability.