High Ecosystem Stability Under Drought Events in National Nature Reserves in China’s Forest Ecosystem

Abstract

Forest-type national nature reserves and their surrounding areas have experienced a series of drought events, which have influenced forest ecosystem stability. Assuming that drought events do not cause a shift in the ecosystem’s stable state, we quantified the stability of forest ecosystems in China’s national nature reserves and their surrounding areas in response to drought events from 2000 to 2018, using satellite-observed Enhanced Vegetation Index (EVI) and Standardized Precipitation Index (SPI) data. We further examined differences in ecosystem stability across regions and forest types, and identified the impacts of environmental factors using correlation analysis, analysis of variance (ANOVA), and random forest models. The results show that both national nature reserves and their surrounding areas primarily experienced single, moderate-intensity drought events, most of which occurred in spring and summer. Compared with surrounding areas, national nature reserves exhibited higher ecosystem stability, with a mean drought resistance index of 31.45 ± 21.09. The difference in ecosystem stability between reserves and their surrounding areas was most pronounced in deciduous forests, which was attributed to their high hydraulic conductivity and distinctive leaf phenological traits. Additionally, climatic factors were the dominant drivers of both resistance and recovery rate, each contributing more than 30% to the overall explained variance. Our results provide valuable guidance for enhancing drought resilience and promoting the sustainable management of China’s national forest reserves.

1. Introduction

Drought events, defined as a precipitation deficiency over an extended period of time, lead to water shortages [1]. Both climate change—especially, rising average temperatures driven by human-generated emissions of heat-trapping greenhouse gases—and regional climate conditions can exacerbate the occurrence of drought events [2,3]. In recent years, drought events have shown an increasing trend [4,5]. Drought plays a critical role in shaping ecosystems [6], significantly affecting the composition, structure, and function [7,8]. These impacts include accelerated leaf senescence, as well as reduced photosynthesis and carbon absorption [9,10,11]. The ecosystem response to drought events, that is, the ecosystem’s ability to persist and recover from drought events, is commonly characterized by two key metrics: resistance and recovery rate [12,13]. Resistance is defined as the percentage of ecosystem loss during drought events relative to its state before drought [14,15]. Recovery rate, by contrast, refers to the percentage of drought-induced loss that is recovered within one year after the drought [16,17]. With ongoing climate change and global warming, drought events are expected to become more frequent and intense, thereby increasing their impacts on ecosystems [6]. Therefore, exploring how ecosystems respond to drought events is crucial for safeguarding ecological security and promoting sustainable development.

Drought events pose a threat to many countries and regions worldwide [18]. National nature reserves are defined as a certain geographical area designated to protect the natural environment and resources, as well as representative natural landscapes and ecosystems [19]. The establishment of national nature reserves is the most fundamental approach to safeguarding the ecological environment and ensuring ecological security [20,21]. China has established a series of national nature reserves to protect diverse ecosystems and wildlife species [22]. By 2016, the number of national nature reserves had increased to 446. A growing body of research indicates that the national nature reserves are affected by global climate change, such as drought events [23]. These impacts include habitat destruction, biodiversity decline, degradation of ecosystem services, and disruptions to community livelihoods [21]. The ecological environment and security of national nature reserves have attracted global attention [24]. Some studies have suggested that drought events may threaten the effectiveness of national nature reserves in the future [25]. Due to shifts in the distribution of protected species, national nature reserves could lose their protective function [26]. While previous research has questioned the future effectiveness of reserves under climate change, few studies have quantitatively assessed how protected areas respond to drought events. The matching method offers a robust way to evaluate such differences [27]. This method compares specific characteristics within protected areas with those in unprotected areas or buffer zones. By comparing ecosystems and species inside and outside national nature reserves, scholars have confirmed that these reserves are effective in maintaining forest cover [28]. However, most reserves in China were established without accounting for climate change factors, such as drought [26], highlighting the need to investigate how these reserves respond to drought events in order to enhance their resilience. Despite its importance, knowledge about the drought response of national nature reserves remains limited.

The response of national nature reserves to drought events is driven by multiple factors. Forest ecosystems are the primary protected objects in national nature reserves; in China, there are 174 forest-type national nature reserves accounting for 46% of the total number of national nature reserves [29]. Under drought conditions, China’s forest ecosystems are among the most severely impacted regions globally [26]. Studies based on the Enhanced Vegetation Index (EVI) have shown that evergreen broadleaf forests (EBF) exhibit stronger resistance than deciduous forests under severe drought events [30]. However, deciduous forests demonstrate higher resistance than evergreen forests during moderate drought events [31]. This highlights the critical role of drought intensity in shaping the response of forest ecosystems to drought. Beyond drought intensity, other factors also influence how forest ecosystems respond to drought events [32]. For example, drought events can reduce species diversity [33] and increase temperature and solar radiation [30]. Therefore, understanding the driving factors behind the response of forest ecosystems to drought events is crucial for enhancing the effectiveness of forest-type national nature reserves [34]. Nevertheless, a significant research gap remains in comprehensively understanding how national nature reserves respond to drought events and the factors underlying these responses, which currently limits efforts to improve their capacity to mitigate climate change impacts.

To address the gap, this study used Standardized Precipitation Index (SPI) data to identify drought events, and EVI data to quantify the response of China’s forest ecosystems to drought from 2000 to 2018. It also identified the driving factors behind these ecosystem responses to drought. The research was conducted in three main steps. First, we analyzed the characteristics of drought events, including their spatial distribution, intensity, frequency, and timing. Second, we examined the spatial patterns of resistance and recovery rate of China’s forest ecosystems. Using the matching method, we further explored differences in resistance and recovery rate between the national nature reserves and their surrounding areas, as well as between the evergreen and deciduous forests. Third, we used random forest regression to identify the key drivers shaping the spatial patterns of resistance and recovery rates in China’s forest ecosystems. Specifically, this study aimed to address the following research questions:

RQ1. 

What are the drought characteristics (e.g., intensity, frequency, timing) of national nature reserves and their surrounding areas?

RQ2. 

Are national nature reserves more resilient to drought events than their surrounding areas?

RQ3. 

Do climatic factors play a dominant role in driving differences in drought response between national nature reserves and their surrounding areas?

2. Materials and Methods

2.1. Study Region

This study focused on 174 national nature reserves classified as ‘forest ecology’ type (Table S1), selected from the National Nature Reserve List published by the Ministry of Ecology and Environment of the People’s Republic of China (http://www.mee.gov.cn, accessed on 18 October 2025), and these nature reserves were classified as ‘forest ecology’ type. The spatial locations and boundaries of these reserves were obtained from the Data Center for Resources and Environmental Sciences, Chinese Academy of Sciences (http://www.resdc.cn, accessed on 18 October 2025).

2.2. Dataset

The datasets collected for this study included land cover and land use data, Enhanced Vegetation Index (EVI) data, meteorological data, and other relevant variables (

Table 1. Data summary in this study.

Data	Period of Time	Temporal Resolution	Spatial Resolution	Data Source
Land cover and land use	2015	/	1 km	Wu et al. (2014) [35]
Enhanced Vegetation Index (EVI)	2000–2018	16-day	1 km	NASA Land Processes Distributed Active Archive Center
Climatic factors—temperature	2000–2018	8-day	1 km	National Ecosystem Science Data Center
Climatic factors—precipitation	2000–2018	8-day	1 km	National Ecosystem Science Data Center
Climatic factors—global radiation	2000–2018	8-day	1 km	Ren et al. (2018) [36]
Stand age	2005	/	1 km	Zhang et al. (2014) [37]
Tree density	/	/	1 km	Crowther et al. (2015) [38]

). Their specific applications were as follows: Land cover and land use data were used to distinguish between evergreen and deciduous forests, enabling analysis of their differential responses to drought events. EVI data served as a proxy for canopy “greenness” and was used to calculate drought resistance and recovery rates. Meteorological data were used to analyze factors influencing ecosystem resistance and recovery, with precipitation data specifically employed to calculate the Standardized Precipitation Index (SPI) as a drought characterization metric. Other variables, including stand age and tree density, were used to investigate additional factors affecting the resistance and recovery of forest ecosystems under drought conditions.

2.2.1. Land Cover and Land Use

The land cover and land use data used in this study, with a spatial resolution of 1 km, were obtained from Wu et al. (2014) and corresponded to the 2015 period [35]. Forest ecosystems dominated the study region, including five primary types: evergreen broadleaf forest (EBF), evergreen needleleaf forest (ENF), deciduous broadleaf forest (DBF), deciduous needleleaf forest (DNF), and mixed forest (MF). To analyze the differential responses of evergreen and deciduous forests to drought events, we grouped the forest types as follows: evergreen forest: combined EBF and ENF, deciduous forest: combined DBF and DNF (Figure 1). Mixed forest was excluded from the analysis due to its small area and complex structural characteristics.

2.2.2. Enhanced Vegetation Index (EVI) Data

The Moderate Resolution Imaging Spectroradiometer (MODIS) EVI products (MOD13A2) were obtained from the NASA Land Processes Distributed Active Archive Center (http://lpdaac.usgs.gov, accessed on 18 October 2025). The data have a spatial resolution of 1 km and a temporal resolution of 16 days, allowing the capture of short-term changes in vegetation growth, which is critical for monitoring drought impacts. Meanwhile, EVI has been widely used as a proxy of canopy ‘greenness’ [39,40], and was specifically developed to address potential saturation issues of traditional vegetation indices and to enhance sensitivity in high-biomass regions [41]. The calculation of EVI was as follows:

$$EVI=2.5\times {\displaystyle \frac{{\rho }_{nir}-{\rho }_{red}}{{\rho }_{nir}+6\times {\rho }_{red}-7.5\times {\rho }_{blue}+1}}$$

(1)

where ${\rho }_{nir},{\rho }_{red}$, and ${\rho }_{blue}$ represent the near-infrared band (841–876 nm), red wave band (620–670 nm), and blue wave band (459–479 nm) of the MODIS sensor (Lockheed Martin Space Systems Company, Bethesda, MD, USA).

To examine how water balance affects forest ecosystems, we first processed the 16-day temporal resolution EVI data using the maximum value composite (MVC) method to generate annual-scale EVI datasets [42]. This step minimizes the interference of cloud cover, atmospheric aerosols, and other noise, ensuring the reliability of EVI-derived metrics for subsequent analysis related to water balance and drought impacts.

2.2.3. Climatic Factors

In this study, temperature and precipitation data, with a spatial resolution of 1 km and an 8-day temporal resolution, were obtained from the published dataset of the National Ecosystem Science Data Center (NESDC) (https://www.nesdc.org.cn/sdo/detail?id=6139dbe27e28173dfbf05947, accessed on 18 October 2025) [43]. Global radiation data were sourced from Ren et al. (2018) [36]. Specifically, they calculated global radiation using field-observed sunshine duration data and the Angstrom model [44]. The resulting site-scale global radiation data were then interpolated to a 1 km spatial resolution and 8-day temporal resolution grid using ANUSPLIN 4.2, with digital elevation model (DEM) data incorporated to optimize the fitting accuracy.

Based on the 8-day scale temperature, precipitation, and global radiation data, we derived three annual-scale metrics: mean annual temperature (MAT), mean annual precipitation (MAP), and mean annual global radiation (MAR).

2.2.4. Other Auxiliary Data

Stand age data were obtained from Zhang et al. (2014) [37]. This dataset was generated using 2005 remote sensing-derived tree height data, combined with the age–height relationship retrieved from field observations. Tree density data with 1 km spatial resolution were sourced from Crowther et al. (2015) [38]. They calculated tree density using a predictive regression model, with model inputs including climate, terrain, vegetation characteristics, and human land use features. Notably, this tree density data specifically reflects the competitive pressure among individual trees with a diameter at breast height greater than 10 cm.

2.3. Methods

Figure 2 shows the flowchart of this study. First, we applied the matching method to define the surrounding areas of national nature reserves; subsequent analyses were then conducted with the national nature reserves and their surrounding areas as the study region. The specific analysis steps were as follows: we identified the occurrence of drought events and quantified key drought characteristics, including drought-affected area, drought frequency, drought severity, and drought timing. We computed the resistance and recovery rate of forest ecosystems in response to drought events. We investigated the factors influencing drought resistance and recovery rate using a random forest (RF) regression model and correlation analysis. Additionally, we used analysis of variance (ANOVA) to test for significant differences in drought resistance and recovery rate between two pairs of groups:

(1)

national nature reserves and their surrounding areas;

(2)

evergreen forests and deciduous forests.

2.3.1. Matching Method

The matching method is a key approach for investigating differences between national nature reserves and their surrounding areas [45]. A critical prerequisite for determining the comparison space (i.e., surrounding areas) is ensuring the relative consistency of vegetation types and natural environmental background conditions [46].

Based on the principle of natural environment heterogeneity and the spatial proximity effect of national nature reserves, we used buffer analysis to define the surrounding areas. Specifically, a 30 km buffer zone was established around each reserve, following previous ecological comparison studies [47,48,49], to represent the surrounding areas while minimizing environmental heterogeneity. This 30 km circular buffer zone was then defined as the surrounding areas of national nature reserves.

2.3.2. Identification of Drought Events

SPI is a drought indicator reflecting long-term water balance, calculated using precipitation data [50,51]. Previous studies have shown that SPI and SPEI (Standardized Precipitation Evapotranspiration Index) are consistent in humid and semi-humid areas but differ significantly in arid and semi-arid areas [52]. Since our study area—China’s forest ecosystems—is mostly located in humid and semi-humid areas, SPI exhibits good applicability here. The calculation of SPI followed three key steps: firstly, 8-day precipitation data was integrated to the monthly scale. The monthly precipitation data were further aggregated over timescales ranging from 3 to 24 months [50]. After the above aggregation schemes, the data were normalized using a three-parameter log-logistic distribution; the resulting value is the SPI [51]. For instance, SPI 12 represents the cumulative water balance over the preceding 12 months. A value of 0 indicates average moisture conditions, positive values indicate humid conditions, and negative values indicate dry conditions.

We used SPI 12 to assess the impact of annual water balance on forest ecosystems, for two reasons:

(1)

longer-timescale SPI is more sensitive to identifying hydrological drought [53];

(2)

plant growth is regulated by accumulated precipitation over the past 12 months [54].

Furthermore, monthly SPI 12 data were integrated into the annual scale. Drought severity was classified based on SPI-12 thresholds: moderate drought: SPI-12 between −1 and −1.5 [31]; severe drought: SPI-12 below −1.5 [31]. We also calculated two additional drought metrics using SPI-3. We summarized at the pixel scale based on the spatial extent of drought events as drought occurrence frequency. Drought timing was also determined by seasonal SPI-3 values. When SPI 3 data were less than −1 in March, April, or May, spring droughts occurred. Summer droughts were indicated in June, July, or August when the SPI 3 data is lower than −1. The arrival of autumn drought is determined by SPI 3 data less than −1 in September, October, or November. In December of the previous year, January, or February, winter droughts are indicated by SPI 3 data less than −1.

2.3.3. Calculation of Ecosystem Stability

Based on the drought events identified using SPI-12, we calculated the resistance of forest ecosystems to drought events and the recovery rate of forest ecosystems after drought events. Two components of ecosystem stability, including resistance and recovery rate, were widely used to evaluate how forest ecosystems respond to climate extremes. Specifically, resistance reflects the degree to which forest ecosystems are affected by drought disturbances, while recovery rate represents the speed at which ecosystems recover after disturbances [55]. Both metrics assume that forest ecosystems do not shift to an alternative stable state when facing drought disturbances [56]. The calculation method is as follows:

$$Resistance=\left|{\displaystyle \frac{\overline{Y_n}}{\overline{Y_n}-Y_e}}\right|$$

(2)

$$Recovery rate=\left|{\displaystyle \frac{\overline{Y_n}-Y_e}{\overline{Y_n}-Y_{e+1}}}\right|$$

(3)

where $Y_n$, $Y_e$, and $Y_{e+1}$ represent the EVI during normal years (mean across all the non-extreme years), during the year climate extremes occurred, and during the year after a climate extreme, respectively.

In this study, both resistance and recovery rate are dimensionless metrics, allowing direct comparison across different vegetation types. A higher resistance suggests a smaller reduction in EVI during drought events, whereas a higher recovery rate implies a faster post-drought recovery of EVI.

2.3.4. Analysis of Variance (ANOVA)

In this study, one-way analysis of variance was employed to evaluate differences in the resistance and recovery rates of forest ecosystems to drought events. Two sets of comparisons were conducted: (1) between national nature reserves and their surrounding areas; (2) between evergreen forests and deciduous forests. Statistical significance was defined as a p-value < 0.05.

2.3.5. Correlation Analysis

The Pearson correlation coefficient is widely used to measure the degree of correlation between two variables, with values ranging from −1 to 1. The calculation method is as follows:

$$r={\displaystyle \frac{\sum _{i=1}^n(X_i-\overline{X})(Y_i-\overline{Y})}{\sqrt{\sum _{i=1}^n{(X_i-\overline{X})}^2}\sqrt{\sum _{i=1}^n{(Y_i-\overline{Y})}^2}}}$$

(4)

where $X$ and $Y$ represent the impact factors (e.g., MAT, MAP, MAR, stand age, and tree density) and the response of forest ecosystems under droughts, respectively. n refers to the research period, and i refers to the specific year.

2.3.6. Random Forest Regression Analysis

To assess the relative importance of multiple environmental factors on the resistance and recovery rate of forest ecosystems to drought events, this study employed a random forest regression [57]. The predictor variables included MAT, MAP, MAR, stand age, and tree density as the predictor variables. This method has been widely adopted primarily for two reasons. First, it effectively addresses multicollinearity among variables. Second, it can capture nonlinear relationships between predictor variables [58]. The importance of each predictor variable was quantified using the percentage increase in mean square error (%IncMSE) between observations and predictions. This metric reflects the percentage decline in model accuracy when a specific predictor variable is randomly permuted [59].

For the calculation of importance measures, we first computed values for each individual tree, then averaged these values across the entire forest (encompassing 500 trees). To mitigate overfitting, the maximum depth of the decision tree was set to 10. The model’s performance was evaluated using two metrics: mean squared residuals (MSE) of the random forest regression model and p-values of the predictor variables.

3. Results

3.1. Characters of Drought Events in National Nature Reserves and Their Surrounding Areas

Overall, 16.44% of forest-type national nature reserves were affected by droughts, compared to 23.31% of their surrounding areas. Within the forest-type national nature reserves, 44.96% of the evergreen forests and 41.94% of the deciduous forests experienced drought events. Outside the forest-type national nature reserves, 52.44% of the evergreen forests and 37.27% of the deciduous forests experienced drought events. The surrounding areas of forest-type national nature reserves were more affected by drought events than the national nature reserves. Evergreen forests had a higher proportion of area affected by drought events compared to deciduous forests.

From 2000 to 2018, 88 out of 174 forest-type national nature reserves in China experienced drought events. Among them, 55 forest-type national nature reserves were located in evergreen forests, and 33 forest-type national nature reserves were located in deciduous forests (

Table 2. Occurrence of drought events in national nature reserves of China’s forest ecosystem.

Forest Type	Evergreen Forests	Deciduous Forests
Forest-type national nature reserve	Badagongshan, Bamianshan, Baimaxueshan, Baishuihe, Baiyunshan, Chayucibagou, Daiyunshan, Duheyuan, Fanjingshan, Fengyangshan-baishanzu, Fujianwuyishan, Gahai-zecha, Gansulianhuashan, Gansuqilianshan, Ganjiangyuan, Gaoligongshan, Goawangjie, Heizhugou, Hupingshan, Huangsang, Jigongshan, Jiangximatoushan, Jiemuxi, Jinzhaitianma, Jinyunshan, Jinggangshan, Junzifeng, Kuankuishui, Kunyushan, Leigongshan, Liankangshan, Longqishan, Longxi-hongkou, Lushan, Mangdangshan, Maoershan, Minjiangyuan, Mulinzi, Nanyuehengshan, Neimengguhelanshan, Qiyunshan, Shennongjia, Taohe, Wuyanling, Wufenghouhe, Wulipo, Xishuizhongyaredaichanglvkuoyelin, Xiongjianghuangchulin, Xuebaoshan, Yading, Yanlingtaoyuandong, Yaoshan, Yaoluoping, Yingzuijie, Changningzhuhai	Baxianshan, Baihuashan, Baotianman, Beijingsongshan, Dabashan, Dahaituo, Daheishan, Daxinganlinghanma, Funiushan, Gaogesitaihanwula, Gurigesitai, Haitangshan, Heibeiwulingshan, Heichashan, Hengrenlaotudingzi, Huzhong, Huaeshan, Hualongshan, Lishan, Liupanshan, Micangshan, Nvluerhushan, Pingheliang, Qingyazhai, Saihanwula, Shanximicangshan, Shanxiziwuling, Shengshan, Tuoliang, Wangqing, Xiaobeihu, Xiaoqinling, Xiaowutaishan

). Most drought-affected reserves (78 out of 88) experienced only one drought event. Only 10 forest-type national nature reserves experienced two or three drought events (Dabashan, Ganjiangyuan, Gurigesitai, Heizhugou, Hualongshan, Jigongshan, Leigongshan, Saihanwula, Wulipo, and Xuebaoshan; Figure 3c). For drought intensity, moderate drought events influenced most of the forest-type national nature reserves in China. Only eight forest-type national nature reserves were affected by severe drought events (Dayunshan, Fujianwuyishan, Ganjiangyuan, Junzifeng, Mangdangshan, Minjiangyuan, Wuyanling, and Xiongjianghuangchulin). In evergreen forests, 47 national nature reserves occurred during moderate drought events, while eight national nature reserves occurred during severe drought events. All nature reserves in deciduous forests have experienced moderate drought events (Figure 3d). Additionally, droughts in spring and summer affected 73 forest-type national nature reserves, while winter droughts affected only 15. For evergreen forests, 25, 20, and 10 national nature reserves experienced drought events in spring, summer, and winter, respectively. For deciduous forests, 7, 21, and 7 national nature reserves were affected by drought events during spring, summer, and winter, respectively (Figure 3e). Notably, drought conditions (intensity and seasonality) in surrounding areas were similar to those inside the reserves.

Over the past two decades, China’s forest-type national nature reserves and their surrounding areas primarily experienced single drought events. Most droughts were of moderate intensity and occurred predominantly in spring and summer.

3.2. Ecosystem Stability to Drought Events in National Nature Reserves and Their Surrounding Areas

In China, the average resistance and recovery rates of forest-type national nature reserves to drought events showed high spatial variability (Figure 4a,b). The average resistance index in forest-type national nature reserves (31.45 ± 21.09) was significantly higher than in the surrounding area (26.37 ± 14.94), indicating stronger drought tolerance (Figure 4c). After drought events, the recovery rates of forest-type national nature reserves and surrounding areas were 1.30 ± 1.06 and 1.36 ± 1.08, respectively (Figure 4d). This indicated that forest-type national nature reserves have higher resistance to drought events compared with surrounding areas (p < 0.05;

Table 3. Differences in ecosystem stability to drought events between national nature reserves and surrounding areas.

Resistance to Drought Events	National Nature Reserves	The Surrounding Areas
National nature reserves	/	0.001
The surrounding areas	0.003	/
Recovery rate after drought events	National nature reserves	The surrounding areas
National nature reserves	/	0.001
The surrounding areas	0.002	/

Note: The values are the results of one-way ANOVA (p-values).

). However, there was no notable difference in post-drought recovery rates between the reserves and their surrounding areas.

Evergreen and deciduous forests exhibited distinct drought responses, reflecting their contrasting hydraulic and phenological adaptations. In evergreen forests, higher resistance to drought events was observed in the forest-type national nature reserves, when compared with the surrounding areas, with a difference of 16.91% (p < 0.05; Figure 4e;

Table 4. Differences in ecosystem stability to drought events between different forest types at national nature reserves and surrounding areas.

Resistance to Drought Events	National Nature Reserves—Evergreen Forests	The Surrounding Areas—Evergreen Forests	National Nature Reserves—Deciduous Forests	The Surrounding Areas—Deciduous Forests
National nature reserves-Evergreen forests	/	0.001	0.001	0.001
The surrounding areas-Evergreen forests	0.001	/	0.001	0.001
National nature reserves-Deciduous forests	0.001	0.001	/	0.001
The surrounding areas-Deciduous forests	0.001	0.001	0.001	/
Recovery Rate After Drought Events	National Nature Reserves-Evergreen Forests	The Surrounding Areas-Evergreen Forests	National Nature Reserves-Deciduous Forests	The Surrounding Areas-Deciduous Forests
National nature reserves-Evergreen forests	/	0.853	0.002	0.001
The surrounding areas-Evergreen forests	0.853	/	0.001	0.001
National nature reserves-Deciduous forests	0.002	0.001	/	0.001
The surrounding areas-Deciduous forests	0.001	0.001	0.001	/

Note: The values are the results of one-way ANOVA (p-values).

). The difference in recovery rate after drought events between the forest-type national nature reserves and surrounding areas was relatively small, at a small difference of −1.16% (p = 0.853; Figure 4f and Table 4). In deciduous forests, the forest-type national nature reserves’ resistance to drought events was higher than the surrounding areas, with a large difference of 21.13% (p < 0.05; Figure 4e and Table 4). In contrast to evergreen forests, forest-type national nature reserves had significantly higher post-drought recovery rates than surrounding areas (p < 0.05; Figure 4e and Table 4).

Table 5. Correlation of environmental factors and ecosystem stability under droughts.

Pearson Correlation Coefficient	Resistance to Drought Events		Recovery Rate After Drought Events
	National Nature Reserves	The Surrounding Areas	National Nature Reserves	The Surrounding Areas
MAR	0.056 *	0.077 *	−0.049 *	−0.107 *
MAT	0.154 *	0.069 *	−0.017	−0.045
MAP	0.037 *	0.054 *	−0.031 *	−0.15 *
Stand age	0.025 *	0.009	−0.002	−0.064
Tree density	0.007	0.006	−0.003	−0.107 *

Note: * means that p < 0.05. The degree of freedom is 6264.

shows that significant differences in drought resistance between the forest-type national nature reserves and surrounding areas were mainly located in the deciduous forests. ANOVA results confirmed that there was no significant difference in post-drought recovery rates between reserves and surrounding areas in evergreen forests (Table 4).

3.3. Attribution of Ecosystem Stability to Drought Events in Nature Reserves and Surrounding Areas

Climatic factors dominated the drought resistance and post-drought recovery rates of forest ecosystems, compared to stand age and tree density (Figure 5). Below is a detailed analysis of their relative importance and correlation with the two key metrics (drought resistance and recovery rate). Among these, MAR and MAT were the primary factors influencing drought resistance in both forest-type national nature reserves and surrounding areas, with %IncMSE values of 44.27, 58.28, 38.75, and 51.76%, respectively. Tree density, mean annual precipitation (MAP), and stand age were secondary influencing factors (Figure 5a). Environmental factors showed a significant positive correlation with drought resistance in both forest-type national nature reserves and their surrounding areas (Table 5). MAT and MAR had a high correlation with drought resistance, with correlation coefficients of 0.056, 0.154 and 0.077, 0.069 in forest-type national nature reserves and their surrounding areas, respectively.

As illustrated in Figure 4b, the key predictors of post-drought recovery rate differed slightly by location, but climatic factors remained dominant. MAR (%IncMSE = 40.75%) and MAP (%IncMSE = 37.48%) were the most important predictors of the recovery rate after drought events in national nature reserves (Figure 5b). In the surrounding areas, the relative importance of MAR (62.03%) and MAP (53.60%) on the recovery rate after drought events was also large (Figure 5b). MAT, stand age, and tree density were the secondary predictors of the recovery rate of forest ecosystems after drought events. Environmental factors were negatively correlated with post-drought recovery rate in both forest-type national nature reserves and surrounding areas (Table 5). MAR and MAP have a larger correlation with drought recovery rate than other impact factors, with correlation coefficients of −0.049, −0.031 and −0.107, −0.15 in forest-type national nature reserves and their surrounding areas, respectively.

Most forest-type national nature reserves are located in mountainous areas, in contrast to their surrounding area. This geographical difference leads to key environmental and stand structure differences that may enhance drought resistance. The high altitude of these mountainous reserves results in lower temperatures and reduced precipitation relative to surrounding areas (Figure 6c,d). Forest-type national nature reserves also have significantly older stand age and larger tree sizes than their surrounding areas (Figure 6f,g). These combined factors (cooler, drier conditions and more mature forest structure) may explain why forest-type national nature reserves exhibit higher resistance to drought events than their surrounding areas.

4. Discussion

4.1. The Effectiveness of Establishing National Nature Reserves

Forest-type national nature reserves exhibited higher drought resistance than their surrounding areas. This result confirms the significant protective effect of national nature reserves, while also implying that surrounding areas have been subjected to a certain degree of human disturbance. Our findings corroborate earlier research demonstrating that protected areas generally maintain higher ecosystem productivity and resilience under climate stress [20]. For example, Zhang et al. (2016) found that the increase in net primary productivity in most sample areas within the Qinghai–Tibet Plateau Nature Reserve was higher than that in corresponding off-reserve sample areas [60]. De Almeida-Rocha et al. (2021) observed that buffer zones adjacent to protected areas in Brazil were far more degraded than the protected areas themselves [61]. Fan et al. (2025) pointed out that, in the Yellow River Basin of China, the establishment of nature reserves mitigated the decline in gross primary productivity (GPP) within reserves during dry years, compared to off-reserve areas [62]. These consistent results, to some extent, verify the success of nature reserve establishment, supported by relevant policies and strict regional management measures. Our study further provides valuable insights for optimizing such policies and management, with three key recommendations: Firstly, decision-makers should maintain the effectiveness of national nature reserves and formulate guidelines for vegetation protection within reserves [63]. Secondly, ecosystem monitoring in national nature reserves should be strengthened, combining field surveys with high-resolution satellite imagery. Lastly, consider revising local government performance evaluation systems to incorporate ecological capital assessment [64]. To effectively protect critical natural ecological spaces and maintain ecosystem integrity—such as the interconnected ‘mountains, rivers, forests, fields, lakes, and grasslands’ system—it is essential to enhance national ecological security and ensure the stability and sustainable development of nature reserves. This approach will have far-reaching significance for both conservation efforts and broader environmental sustainability.

However, the establishment of nature reserves may inadvertently trigger spillover effects. This spatial displacement of human pressure underscores the need for integrated regional management [65]. This leads to reduced disturbance inside reserves but increased disturbance in surrounding buffer zones [66]. This highlights the importance of implementing comprehensive regional management strategies to mitigate such spillover effects, while also reinforcing the need to protect not only the reserves themselves but also their representative natural landscapes and ecosystems.

In addition, observational data show that drought events became more widespread on the Qinghai–Tibet Plateau and in Southwest and Southeast China during 1980–2015 [67]. Projections from the Coupled Model Intercomparison Project Phase 5 (CMIP5) indicated that under 1.5 °C and 2.0 °C of global warming, the intensity and areal coverage of drought will increase in China [68]. Against this backdrop of projected more frequent droughts, exploring the differences in drought response between national nature reserves and their surrounding areas is crucial for enhancing the drought resilience of national nature reserves [69,70].

4.2. Deciduous Forests Are More Resistant to Drought Events than Evergreen Forests

In terms of forest type, we found that deciduous forests exhibited higher resistance to drought events than evergreen forests. This result aligns with conclusions from studies in karst areas [71] and tropical arid regions [72], reflecting distinct drought adaptation strategies between the two forest types. There were clear differences in the adaptation strategies of the evergreen and deciduous forests under drought conditions. The superior drought resistance of deciduous forests likely arises from their higher hydraulic conductivity and adaptive leaf-shedding strategies that minimize water loss [73]. For instance, carbon allocation prioritizes roots and stems during drought, enhancing water storage capacity in these organs. Leaf abscission acts as a ‘hydraulic fuse’, preventing drought-induced damage to hydraulically vulnerable xylem [74]. Under drought stress, deciduous forests respond with early leaf discoloration and premature senescence—traits that reduce water loss [75]. Evergreen forests retain leaves under drought stress, but this comes with trade-offs: they maintain relatively low photosynthetic capacity and reduced leaf water transport efficiency [76,77]. Year-round leaf retention (with no distinct leaf unfolding or falling dates) increases their exposure to drought events [30]. Drought can trigger xylem cavitation and plant tissue desiccation via hydraulic failure [78,79]. In response, evergreen forests often close stomata to avoid hydraulic damage, which may lead to carbon starvation [73,80]. Notably, despite differences in resistance, the post-drought recovery rates of evergreen and deciduous forests were similar. This aligns with known functional traits: deciduous forests have high growth rates due to rapid resource acquisition [72], while evergreen forests excel at efficient conservation and reuse of internal resources [81]. Further research is needed to collect more data on post-drought recovery rates across different forest types to validate this pattern.

We found that deciduous forests have high drought resistance and similar recovery rates between the two forest types, indicating that mixing evergreen and deciduous forests could be an effective strategy to cope with climate change. This is supported by existing research. Mixed evergreen-deciduous stands enhance stand-level photosynthetic performance, with optimal biomass mixing effects observed at an equal ratio of the two tree types [82]. Mixed forests offer additional benefits, including improved soil fertility, enhanced carbon sequestration, and higher biodiversity [83]. The reduced drought vulnerability of mixed forests can be explained by three mechanisms: resource partitioning, facilitation, and selection effects [83].

4.3. The Dominant Role of Climatic Factors on Forest Ecosystems’ Response to Drought Events

Climatic factors are the primary drivers of ecosystem stability when facing drought events. Our results align with two key global and regional studies. Huang and Xia (2019) emphasized that global radiation and temperature play critical roles in shaping ecosystem responses to drought [30]. Ren et al. (2023) found that environmental factors (e.g., climate) are the main regulators of typical ecosystem stability on the Tibetan Plateau [84]. Climatic factors affect ecosystem stability during and after droughts through multiple pathways. Temperature increases associated with droughts exacerbate water scarcity, reduce water supply and net carbon dioxide exchange, and ultimately lower vegetation productivity [85]. Drought events can also increase solar radiation, which can contribute to a rise in leaf area index [86]—a factor linked to ecosystem water and carbon cycling. The water use efficiency of forest ecosystems during post-drought recovery is a key determinant of their post-drought stability [87].

In addition, stand age and tree density do impact ecosystem stability during droughts, although their effects tend to be less prominent when analyzed at larger spatial scales. Beyond these factors, forest responses to droughts are also influenced by other variables, such as biodiversity. Notably, existing research indicates that biodiversity enhances ecosystem drought resistance only in drought-prone environments [88]—suggesting its role is context-dependent.

4.4. Uncertainties and Future Prospects

Our results could help identify existing issues and propose targeted countermeasures to enhance the stability and sustainable development of forest ecosystems in national nature reserves. However, this study has several limitations that can be addressed in the future.

We only considered the impact of meteorological drought on ecosystem stability, without consideration of soil drought. Existing research has shown that soil moisture exerts a greater influence on vegetation growth than precipitation [89]. Drought stress on forest ecosystems is typically gradual: its intensity increases over time, and the ecosystem’s response also changes dynamically [90]. Future studies could incorporate additional hydrological variables (e.g., soil moisture, groundwater level) to more comprehensively characterize drought events and their ecological impacts. Second, two limitations were associated with the use of the EVI to assess the vegetation responses to drought. Canopy stress in plants may stem from non-drought factors (e.g., pests, nutrient deficiency), which cannot be distinguished using EVI alone [41]. Complex canopy structures can produce large-scale shading effects, which may distort remote sensing-based monitoring results of vegetation greenness [91]. Future studies could integrate diverse remote sensing datasets, such as Solar-Induced Chlorophyll Fluorescence (SIF), kernel Normalized Difference Vegetation Index (kNDVI), and Near-Infrared Reflectance of Vegetation (NIRv), to more accurately characterize vegetation greenness and stress.

5. Conclusions

In this study, we examined forest resistance and resilience in forest-type natural reserves of China using satellite-observed EVI and SPI-based drought index over the past two decades. We found that forest-type national nature reserves in China have primarily experienced single drought events. These droughts were mostly moderate in intensity and occurred mainly in spring and summer. National nature reserves exhibited higher drought resistance than their surrounding areas, though their post-drought recovery rates were comparable. Significant differences in ecosystem stability between national nature reserves and their surrounding areas were mainly concentrated in deciduous forests. Climatic factors were the primary drivers of drought resistance and post-drought recovery rates in both national nature reserves and their surrounding areas. Specifically, environmental factors showed a positive correlation with drought resistance and a negative correlation with recovery rate. This study on the impacts of drought events on ecosystem stability provides valuable insights for enhancing forest management effectiveness, improving ecosystem stability amid intensifying droughts, and promoting the sustainable development of national nature reserves.