The Multiple Impacts of Climate Change and Human Activities on Vegetation Dynamics in Yunnan Province, China

Abstract

Vegetation plays an important role in the hydrological cycle, carbon storage and regional climate. It provides multiple ecosystem services, regulates ecosystem structure and promotes the sustainable and stable development of the earth’s ecosystem. Under the interference of the ever-changing environment, vegetation vulnerability is increasingly evident. This study focuses on Yunnan Province, China, where we analyze the spatiotemporal dynamics of NDVI at both provincial and municipal scales. Utilizing methods such as geographical detectors, time-lag analysis, and residual analysis, we identify key drivers of NDVI changes in Yunnan. From 2001 to 2023, the multi-year average NDVI in Yunnan decreases spatially from southwest to southeast, with the annual maximum NDVI increasing at a rate of 0.025 per decade. Qujing City exhibits the fastest NDVI growth, while Diqing City shows the slowest. Vegetation degradation is primarily concentrated in central Yunnan. The NDVI in Yunnan demonstrates significant spatial heterogeneity, influenced by a combination of climatic, topographic, and anthropogenic factors. The interaction between land use type and precipitation is identified as a key driver, explaining over 50% of the spatial distribution of NDVI. Approximately 83% and 82% of vegetated areas in Yunnan exhibit delayed responses to precipitation and temperature changes, respectively. Notably, 73% of the NDVI increase and 7% of the NDVI decrease in Yunnan were jointly affected by climate change and human activities, and positive contributions from these factors cover 92% and 90% of the area, respectively. The impact of human activities on vegetation is mainly positive, although urbanization in central Yunnan significantly inhibits NDVI. By elucidating key mechanisms, this work fosters balanced vegetation–environment synergies in Yunnan and supports the building of ecological safeguards in China.

1. Introduction

Vegetation serves as a critical regulatory mechanism in sustaining Earth’s hydrological cycles, modulating energy fluxes, and maintaining carbon sequestration equilibrium [1,2]. Vegetation not only improves the ecological environment but also contributes to mitigating climate change [3], making it a key indicator for assessing ecosystem health and the effectiveness of ecological restoration [4]. Increased greening can enhance productivity, boost terrestrial carbon storage, slow anthropogenic climate warming, and alter surface biogeophysical properties, thereby influencing the hydrological cycle and climate change across multiple spatial and temporal scales [5]. This has significant implications for ecosystem stability and human well-being [6]. Vegetation information is typically obtained through field observations and remote sensing analysis, the latter providing robust technical support for the monitoring and protection of surface vegetation [7,8]. The Normalized Difference Vegetation Index (NDVI), a widely used and easily accessible remote sensing metric [9], has been extensively applied in practices assessing and measuring changes in surface vegetation [10]. Previous research has effectively employed the NDVI to assess changes in vegetation coverage, phenology, and related characteristics at both global and regional scales [11,12]. Therefore, the NDVI can provide us with crucial information regarding the growth status, cover density, and ecosystem health of surface vegetation.

Vegetation changes are influenced by a multitude of factors including climate, topography, and human activities [13]. These complex factors interact to shape the distribution, growth, and health of vegetation. In terms of climate, the amount and variability of rainfall significantly affect the photosynthetic activity and carbon absorption capacity of vegetation. Observed rainfall over a certain period has a non-negligible impact on vegetation [14]. Temperature influences seed germination and vegetation growth, affecting enzyme activity, respiratory metabolism, photosynthesis, leaf temperature, and photosynthetic efficiency [15]. The mechanism of human activities on vegetation is complex. For example, from 2000 to 2020, China implemented extensive “Grain for Green” policies, which restored 63% of vegetation cover in areas prone to desertification [16]. However, human activities can also have detrimental effects, such as in Africa where population growth has led to agricultural expansion, urbanization, and increased demand for wood fuel, resulting in a decline in natural forests and other woody vegetation cover [17]. In recent years, the trend toward normalization of extreme climate events may have led to long-term reductions in vegetation cover [18]. Human disturbances have increased the vulnerability and risk of vegetation in the context of climate change [19]. Therefore, uncovering the drivers of spatiotemporal vegetation changes is crucial for promoting ecological restoration and conservation.

However, research on vegetation dynamics and their driving mechanisms in complex mountainous environments holds particular significance and urgency, while also presenting substantial challenges. Mountainous areas cover 25% of the world’s land surface yet harbor the planet’s highest concentration of plant endemism [20]. The rich vegetation is an indispensable component for maintaining mountain ecosystem stability [21]. Nevertheless, complex topography renders these regions exceptionally sensitive to climate change, placing ecosystem functionality and resilience at severe risk [22,23]. Studies indicate that even without significant surface temperature increases, mountain biodiversity faces critical threats [24]. Consequently, long-term monitoring of mountain vegetation carries profound theoretical and practical implications. Yunnan Province in southwestern China exemplifies such mountainous terrain, characterized by highly complex and fragmented topography [25]. Shaped by the Himalayan orogeny and intense neotectonic activity, its eco-environment exhibits pronounced regional heterogeneity, biodiversity, and vulnerability [26]. Climate patterns are exceptionally complex and spatially heterogeneous [27]. Vegetation changes here result from intertwined natural and anthropogenic factors with marked spatial variability.

The geographical detector method excels in simultaneously analyzing numerical and qualitative data [28], not only identifying the main factors affecting dependent variables but also revealing the complex interdependencies among socioeconomic, geographical, and climatic factors [29]. This makes it a robust tool for investigating spatial heterogeneity in geographical phenomena [30]. Given Yunnan’s climatic complexity, conventional linear and correlation analyses fail to capture intricate vegetation–climate–human activity relationships [8]. Therefore, further exploring the nonlinear impact of climate factors on vegetation change and the time lag effect of vegetation response to climate change and quantitatively distinguishing the relative contribution of climate change and human activities to vegetation dynamics will help to fully establish the complex relationship between socioeconomic development, atmospheric conditions and plant growth. Earlier studies attempted to reflect regional vegetation conditions by dividing the province into secondary scales such as central, eastern, southern, and western regions [31,32,33], but this approach overlooks critical intra-provincial variations. Complex topography fosters localized microclimates—for instance, Xishuangbanna’s low-latitude valley position sustains a perennial warm-and-rainy tropical rainforest climate distinct from surrounding regions. Such oversimplification likely masks significant differences in vegetation growth dynamics and influencing factors across prefectural boundaries.

In summary, clarifying the spatiotemporal pattern of vegetation in Yunnan Province, identifying the main driving factors affecting vegetation change and their complex interactions, and distinguishing and measuring the impacts of climate change and human intervention are important foundations and prerequisites for protecting regional ecosystems and sustainable development. Therefore, this study focuses on Yunnan Province and its 16 prefecture-level cities, aiming to (1) analyze the trend of NDVI from 2001 to 2023; (2) evaluate how topography, climate variables, and human activities shape the spatial heterogeneity of NDVI; (3) reveal the delayed response of NDVI to precipitation and temperature changes; and (4) quantify the contribution of climate and anthropogenic factors to NDVI changes. The findings provide scientific support for ecological environmental protection and construction in Yunnan Province and its subregions at different temporal and spatial scales, ensuring long-term environmental resilience.

2. Materials and Methods

2.1. Study Region

Yunnan Province is located in the southwestern region of China (21°8′~29°15′ N, 97°31′~106°11′ E) (Figure 1a), comprising 8 prefecture-level cities and 8 autonomous prefectures (Figure 1b). The terrain is predominantly mountainous plateau, with higher elevations in the northwest and lower in the southeast, featuring an altitude difference of up to 1000 m [34]. The annual precipitation is 1100 mm [35], characterized by uneven spatial and temporal distribution [36], with distinct wet and dry seasons [37,38]. Yunnan experiences a typical subtropical plateau monsoon climate, with significant climatic variations across different regions due to topographical differences [39]. Yunnan Province stands as a critical biodiversity hotspot both for China and globally, boasting extraordinary botanical richness with over 19,000 plant species—a figure representing more than half of China’s total plant diversity [40]. Its robust ecosystems play a pivotal role in safeguarding ecological security across South and Southeast Asia while delivering essential ecosystem services [41,42,43]. The province is undergoing a warming and drying trend, with a particularly noticeable reduction in precipitation [44], and has experienced frequent heatwaves and high-temperature events in recent years [45]. During 2009–2012 alone, extreme consecutive droughts caused river flows to plummet by 30–80%, triggering widespread desiccation of streams, reservoirs, and ponds, with severe consequences including drinking water shortages and crop failures [46]. Under global warming, Yunnan also faces acute soil erosion issues. According to the Yunnan Climate Change Adaptation Action Plan, karst landscapes cover 28.1% of its territory, while soil erosion affects ~30% of the land area, placing natural habitats (e.g., tropical rainforests, alpine vegetation) at risk of structural degradation, intensified rocky desertification, and species loss. As a crucial hub for China’s economic, cultural, and political exchanges with the outside world, rapid economic development and large-scale afforestation and reforestation programs have significantly impacted land use in Yunnan Province [47]. Specifically, the northwest and southwest serve as ecological barriers and biodiversity hotspots, characterized by high forest coverage and low human activity intensity, though montane agriculture may impact natural vegetation. A high-intensity human activity zone extends from the central to northeastern and southeastern regions, dominated by urban expansion, agricultural intensification, and industrial agglomeration [41,48]. Therefore, monitoring vegetation dynamics and analyzing key driving factors in Yunnan Province are particularly important against the backdrop of rapid climate change and social development.

2.2. Data

The monthly NDVI data (2001–2023) are from the National Tibetan Plateau/Third Pole Environment Data Center (https://data.tpdc.ac.cn/, accessed on 1 December 2024), with a spatial resolution of 250 m and noise removal and time series filtering to ensure accuracy. Annual NDVI datasets were generated using the maximum value composite (MVC) approach.

The photosynthetic carbon fixation process of vegetation is significantly restricted by climatic conditions and has a time lag effect [49]. Vegetation is particularly sensitive to precipitation and temperature [50,51], and the influence of precipitation on vegetation sensitivity has become increasingly prominent over time [52]. At the same time, the thresholds of precipitation and temperature define the spatial distribution of vegetation and form ecological constraints on the relative advantages of different types of vegetation [53]. The annual precipitation and annual average temperature data were calculated using the monthly precipitation dataset with a resolution of 1 km from the National Tibetan Plateau / Third Pole Environment Data Center.

In addition, topography is a non-climatic factor closely related to vegetation distribution [54]. For example, vegetation shows different degrees of greening along the altitude gradient [55]. Topography has a regulatory effect on the ecological adaptability of vegetation. The slope and aspect were calculated using DEM data based on ArcGIS 10.7 software. At the same time, the landform types were merged and reclassified, and finally divided into seven categories: plains, mesa, hills, low-relief mountains, moderate-relief mountains, high-relief mountains and extreme-relief mountains.

Regional soil properties also control the spatial pattern of vegetation diversity [56] and are the basic environmental factors for vegetation growth, further strengthening the ecological constraints on vegetation distribution. We merged and reclassified the soil data attributes, and finally divided the soil types into 9 categories: leached soils, semi-luvisols, aridosols, primary soil, semi-hydromorphic soils, hydromorphic soil, anthrosols, alpine soils and ferralsols.

In addition to natural factors, the impact of human disturbance on vegetation cannot be ignored. It is widely present in all parts of the world (except polar and desert regions) [57]. This study selected three indicators, namely land use type, population and night lights, which are directly or indirectly changing vegetation cover and community structure. In summary, we considered dozens of driving factors that may affect NDVI changes, including precipitation, temperature, elevation, slope, aspect, soil type, landform type, land use, night lights, and population. Finally, cropping, resampling, and other operations are performed on all data to obtain the final model input data. The specific description of the data is shown in

Table 1. Details of the data used in this study.

Type	Data Name	Spatial/Temporal Resolution	Time Range	Data Source
Vegetation index	NDVI	250 × 250 m/monthly	2001–2023	https://data.tpdc.ac.cn/
Meteorological factors [58,59,60,61]	Temperature	1 × 1 km/monthly	2001–2023	https://data.tpdc.ac.cn/
	precipitation	1 × 1 km/monthly	2001–2023	https://data.tpdc.ac.cn/
Topographical factors	Elevation	250 × 250 m	-	https://www.resdc.cn/
	Slope	250 × 250 m	-	https://www.resdc.cn/
	Aspect	250 × 250 m	-	https://www.resdc.cn/
Geomorphology factors	Landform type	1 × 1 km	-	https://www.resdc.cn/
Soil factor	Soil type	1 × 1 km	-	https://www.resdc.cn/
Human factors	Population	1 × 1 km/yearly	2001–2023	https://landscan.ornl.gov
	Land use [62]	30 × 30 m/yearly	2023	https://zenodo.org/records/4417810 (accessed on 2 December 2024)
	Nighttime light [63]	500 × 500 m/yearly	2001–2023	http://www.geodata.cn

. The spatial distribution of various driving factors in Yunnan Province is shown in Figure 2.

2.3. Methods

2.3.1. Trend Analysis

As a robust non-parametric method [64], the Theil–Sen median slope estimator is widely used to assess spatiotemporal change trends in vegetation characteristics [65,66,67,68]. The Mann–Kendall (MK) trend test [69,70] demonstrates significant advantages as it does not require assumptions regarding data distribution characteristics or linear relationships [71]. In the context of remote sensing-based vegetation dynamic monitoring, researchers frequently integrate the Theil–Sen median method with MK testing [72]. This integrated analytical framework effectively captures critical insights into vegetation evolution by simultaneously quantifying both the magnitude of trends and their statistical significance. Since areas with an absolute zero NDVI change trend are virtually nonexistent, minor variations caused by natural noise are categorized as “stable.” Therefore, a threshold of 0.0005 was selected [73]. In this study, changing patterns of NDVI can be classified into five categories: significant improvement (slope >= 0.0005, |Z| ≥ 1.96), not significant improvement (slope >= 0.0005, |Z| ≤ 1.96), stable (−0.0005 < slope < 0.0005), not significant degradation (slope <= −0.0005, |Z| ≤ 1.96), and significant degradation (slope <= −0.0005, |Z| ≥ 1.96).

2.3.2. Geographical Detector

The Geographic Detector is an effective tool for analyzing the spatial heterogeneity of variables, capable of identifying driving mechanisms through spatial association characteristics between variables [74]. The Geographic Detector can not only quantify the contribution of explanatory variables to spatial heterogeneity but also evaluate the synergistic effects of multiple factors (enhancing/weakening/independent) [75,76]. Through methodological extensions, the model has evolved into an interdisciplinary application framework spanning multiple research fields [76].

Vegetation dynamics are jointly regulated by multiple environmental factors, including climate, topography, geomorphology, soil, and human activity disturbances. Drawing on previous studies [31,77], we selected ten factors—precipitation, temperature, elevation, slope, aspect, soil type, geomorphic type, land use type, population, and nighttime light—to investigate the spatiotemporal differentiation mechanisms of NDVI in Yunnan Province. The Optimal Parameter-based Geodetector Model (OPGD) was employed, which innovatively integrates spatial data partitioning optimization algorithms to automatically identify optimal discretization schemes (grouping numbers and breakpoint methods) for explanatory variables [78]. The OPGD model was implemented using the “GD” package in R 4.3.1 [79]. The independent variables are ten driving factors, and the dependent variable is the annual maximum NDVI. Its core hypothesis posits that the greater the spatial influence of independent variables on the dependent variable, the more similar their spatial distributions become [80]. Consequently, input independent variables must be discretized into categorical values based on the dependent variable and computed hierarchically [81]. The optimal discretization scheme is evaluated using the highest q-value [82], where the q-statistic is calculated as the ratio of between-group variance to total variance [83]. Using a spatial scale of 1 km to study the spatiotemporal patterns of NDVI in Yunnan Province not only allows for a stable ranking of driving factors but also emphasizes the spatial heterogeneity of different factors [84]. Referring to previous studies [85], classification levels for explanatory variables were constrained to 5–8 categories. The results demonstrate that elevation, slope, temperature, nighttime light, and population indicators all require partitioning into 8 intervals, but the break methods differ. Elevation and temperature use equal interval breaks, slope and population use geometric interval breaks, and nighttime light uses quantile breaks. The optimal discretization scheme for precipitation involves 6 intervals using standard deviation breakpoints (Figure 3a,b).

$$q=1-{\displaystyle \frac{{\sum }_{h=1}^L{N}_h{\delta }_h^2}{N{\delta }^2}}$$

(1)

where the value of q quantifies the predictive power of the factor (range: [0, 1]). A value of q closer to 1 indicates stronger explanatory capability of the factor. h = 1, ⋯, L denotes the stratification of either explanatory or response variables. N_h and N represent the number of units in stratum h and the total units across the entire region, respectively. Additionally, ${\delta }_h^2$ and ${\delta }^2$ denote the variance of values in stratum h and the total variance across the entire region, correspondingly.

2.3.3. Time-Lagged Effects of Meteorological Factors on Vegetation Changes

To evaluate the dependencies between climatic factors (precipitation, temperature) and NDVI across varying time-lag scales, this study implemented a multi-stage statistical modeling approach. Initially, correlation analysis was conducted to characterize the relationships between climatic factors and NDVI [86]. Subsequently, partial correlation analysis was applied to eliminate interference from confounding climatic variables, thereby accurately identifying dominant drivers of vegetation change and their lag effects [87]. This method enables accurate isolation of environmental variables’ independent effects on vegetation growth at a monthly scale [88]. It has been successfully applied to investigate vegetation responses to various climatic factors across global, China-wide, and subregional domains [89,90,91]. Under 0–3-month lag conditions [77,90,92], the maximum response intensity of NDVI to climatic factors and the corresponding lag durations for precipitation and temperature were determined [93], with specific computational equations detailed below.

$$R_{x,y|z}={\displaystyle \frac{r_{xy}-r_{xz}\times r_{yz}}{\sqrt{\left(1-r_{xz}^2\right)\times \left(1-r_{yz}^2\right)}}}$$

(2)

where r_xy, r_xz, and r_yz represent the correlation coefficients between x (dependent variable), y (independent variable), and z (control variable), respectively. R_x,y|z is the partial correlation coefficient between the dependent variable and independent variable for a given control variable.

2.3.4. Residual Analysis

The dynamic evolution of NDVI is jointly driven by climate variability and anthropogenic pressures. The residual analysis framework enables quantitative differentiation of the relative contributions from meteorological forcing and socioeconomic development pressures to vegetation cover changes [94]. It is a widely recognized and relatively robust method [95], which has emerged as a pivotal methodological paradigm in driver attribution analysis and is widely applied in ecosystem dynamics research [96,97,98,99]. The combined application of partial correlation analysis and residual trend analysis effectively resolves the persistent issue of causal ambiguity inherent in traditional correlation approaches [100]. The specific computational equations are detailed below [101]:

$${NDVI}_{CC}=a\times T+b\times P+c$$

(3)

$${NDVI}_{HA}={NDVI}_{obs}-{NDVI}_{CC\#}$$

(4)

where a, b, and c are the coefficients of the regression model, T and P denote average temperature and total precipitation, and NDVI_HA, NDVI_CC and NDVI_obs denote, respectively, the residual of NDVI, observed NDVI and predicted NDVI based on the regression model.

3. Result

3.1. Spatiotemporal Patterns of NDVI

From 2001 to 2023, the annual maximum NDVI for Yunnan Province increased at a rate of 0.025 per decade (Figure 4d), with a multi-year average NDVI of 0.80. The temporal trends of the annual maximum NDVI values varied across the 16 cities in Yunnan Province, but all showed an overall upward trend. The growth rates of NDVI, ranked from highest to lowest, were as follows: QJ (0.036/10a) > WS (0.033/10a) > ZT (0.032/10a) > PE (0.029/10a) > LJ (0.027/10a) > LC (0.026/10a) > CX (0.025/10a) > HH (0.023/10a) > BS (0.022/10a) > NJ (0.022/10a) > DL (0.020/10a) > XSBN (0.019/10a) > KM (0.018/10a) > YX (0.015/10a) > DH (0.014/10a) > DQ (0.012/10a) (Figure 4d). The multi-year average NDVI values for the 16 cities ranged between 0.70 and 0.90, ranked from highest to lowest as follows: XSBN (0.870) > DH (0.855) > PE (0.847) > LC (0.837) > BS (0.824) > NJ (0.816) > ZT (0.813) > WS (0.799) > HH (0.793) > DL (0.784) > YX (0.777) > CX (0.773) > LJ (0.769) > QJ (0.763) > KM (0.748) > DQ (0.743) (Figure 4c).

In total, 98% of the regions in Yunnan Province have an annual average max NDVI > 0.6, indicative of robust vegetation coverage and growth conditions. Spatially, the NDVI values decreased from the southwest to the southeast (Figure 4a). The southwestern region in Yunnan Province (elevation 500–900 m) is dominated by low-mountain topography and abundant precipitation. In some areas, the annual average precipitation may exceed 1600 mm [41], providing optimal hydrothermal conditions sustaining high vegetation density. As shown in Figure 3b, regions where the NDVI trend shows a significant increase and a non-significant increase account for 56% and 28% of the study area, respectively, with a widespread spatial distribution. The process of positive vegetation succession dominates the overall ecological pattern of the study area. Regions with a decreasing NDVI trend account for 9% of the study area (including 3% with a significant decrease). The hotspot regions where NDVI shows a decrease are concentrated in cities such as BS, DL, KM, QJ, YX, and HH. This area spatially overlaps with the core region of the Central Yunnan Urban Agglomeration. With a relatively high level of urbanization, it serves as the core area for urban development across the province. From 2000 to 2020, the impervious surface area within the Central Yunnan Urban Agglomeration increased by 1402 square kilometers, representing a growth rate of 91.5% [102]. Intensive land development is likely the primary cause contributing to the decrease in NDVI in these regions. Rapid urban expansion has encroached on vegetated areas (e.g., forests, grasslands), directly reducing vegetation cover [103]. Furthermore, urban development in central Yunnan radiates outward from its core region, exhibiting a spatial pattern where construction intensity diminishes from southeast to northwest [104]. Consequently, areas of significant vegetation reduction are more prevalent in the southeastern sector of central Yunnan (particularly in the contiguous zones of KM, YX, HH, and QJ) than in the northwest. However, the persistent implementation of major ecological conservation and restoration projects in the rocky desertification belts of eastern and southeastern Yunnan—including comprehensive rocky desertification management programs in QJ, KM, HH, and WS prefectures, along with the ecological rehabilitation project for QJ’s urban-facing mountain areas—has driven an overall upward trend in urban vegetation across central, eastern, and southeastern Yunnan in recent years. The multi-year average NDVI values in northwestern Yunnan were relatively low, with some areas having values between 0 and 0.20. Additionally, the vegetation cover trends in most areas remained largely unchanged, while a few areas showed signs of degradation, all located in DQ. Northwestern Yunnan is adjacent to the Tibetan Plateau and the Sichuan Basin, with elevations reaching up to 6740 m. The annual temperature rise in high-altitude areas is most pronounced [105], resulting in lower vegetation coverage.

3.2. Spatial Heterogeneity Detection of NDVI

According to Figure 5, for the entire Yunnan Province, the explanatory power of each driving factor on the spatial heterogeneity of NDVI, based on the q statistic, is ranked in the following order: land use type > precipitation > landform type > soil type > elevation > temperature > nighttime light > population > slope > aspect. Land use has the greatest impact on NDVI (q = 0.401), with an explanatory power of over 40%. Aspect has the least explanatory power on NDVI (q = 0.002), while the q values of the remaining driving factors range between 0.1 and 0.2. The NDVI in Yunnan Province is influenced by a combination of environmental, social, and economic factors. For example, urbanized areas in central Yunnan Province (such as construction and transportation land) typically lead to a decrease in NDVI values, whereas forested areas in the southwestern region, with high coverage, generally exhibit higher NDVI values. Additionally, NDVI is sensitive to precipitation, which directly affects the spatial pattern of vegetation [106]. Different landform features, such as mountains, plains, or hills, may result in varying hydrological and microclimatic conditions, thereby influencing vegetation growth. Soil type, elevation, and temperature also carry certain weight in the overall analysis. Vegetation growth and rainwater reuse efficiency are significantly influenced by soil type [107]. Changes in topographic attributes such as elevation and slope are often accompanied by variations in climatic and meteorological conditions, which can regulate vegetation greenness and are considered stable controlling factors for vegetation distribution and development [108]. Temperature can promote plant growth and activity [109].

Consistent with the provincial-scale results, the type of land use ranks first in explaining the spatial heterogeneity of NDVI in fourteen cities: BS, CX, DL, DH, DQ, LC, LJ, NJ, PE, QJ, WS, XSBN, YX, and ZT (

Table A1. Factor detection for NDVI in 16 cities of the Yunnan province.

City	Factor Detection Results
BS	Land use type > Population > Nighttime light > Slope > Soil type > Elevation > Landform type > Temperature > Precipitation
CX	Land use type > Temperature > Precipitation > Elevation > Soil type > Landform type > Population > Slope > Nighttime light > Aspect
DL	Land use type > Population > Temperature > Elevation > Landform type > Soil type > Nighttime light > Slope > Precipitation
DH	Land use type > Population > Slope > Elevation > Temperature > Nighttime light > Landform type > Soil type > Precipitation
DQ	Land use type > Temperature > Elevation > Soil type > Landform type > Precipitation > Population > Aspect > Slope
HH	Landform type > Land use type > Slop > Precipitation > Nighttime light > Population > Soil type > Elevation > Temperature
KM	Nighttime light > Land use type > Population > Landform type > Slope > Elevation > Temperature > Soil type > Precipitation
LC	Land use type > Population > Nighttime light > Slope > Precipitation > Landform type > Soil type >Temperature > Elevation > Aspect
LJ	Land use type > Temperature > Elevation > Soil type > Population > Slope > Nighttime light > Landform type > Aspect > Precipitation
NJ	Land use type > Elevation > Temperature > Precipitation > Soil type > Landform type > Population > Slope > Aspect
PE	Land use type > Population > Nighttime light > Slope > Landform type > Soil type > Temperature > Elevation > Precipitation > Aspect
QJ	Land use type > Nighttime light > Landform type > Population > Slope > Precipitation > Temperature > Elevation > Soil type
WS	Land use type > Landform type > Slope > Population > Nighttime light > Elevation > Temperature > Soil type > Precipitation
XSBN	Land use type > Slope > Population > Landform type > Nighttime light > Soil type > Elevation > Temperature > Precipitation
YX	Land use type > Population > Landform type > Nighttime light > Slope > Precipitation > Soil type > Elevation > Temperature
ZT	Land use type > Temperature > Elevation > Nighttime light > Precipitation > Landform type > Slope > Population > Soil type > Aspect

). However, geographical and environmental factors affect NDVI with varying intensity among cities. Cities in central and southwestern Yunnan are primarily driven by non-climatic factors (Figure 5, Table A1). In central Yunnan, changes are predominantly induced by intensive human activities. In KM, urban expansion tripled during 2000–2015 [110]. With high population concentration and rapid economic development, the city is expanding, resulting in lower NDVI values. Areas with fewer residents often have richer natural vegetation and higher NDVI values. For example, in HH, NDVI also correlates strongly with landform type, indicating that the city is located in an area with significant landform differences, such as mountainous or valley areas, where terrain changes may lead to local hydrological and climatic conditions affecting vegetation distribution. For instance, protected areas like the Ailao Mountains spanning CX and HH feature high-elevation topography with rich vegetation and dense forest cover, displaying distinct vertical zonation patterns. Southwestern Yunnan’s NDVI patterns are primarily attributable to its unique geographical endowments. Abundant vegetation and robust ecological resilience make spatial heterogeneity more susceptible to anthropogenic impacts [111]. For northwestern cities like NJ and DQ, NDVI is closely related to altitude (Table A1), likely because these cities are situated in a mountainous area, with some regions located on plateaus, where elevation directly influences climate and vegetation distribution, thereby affecting NDVI. Exemplifying this, the Gongshan Nature Reserve in NJ develops conspicuous forest stratification patterns shaped by its dramatic topography and three-dimensional microclimates.

In summary, land use type dominates NDVI spatial heterogeneity across Yunnan Province, with inter-city differences attributable to varying environmental contexts and development stages. The results of the factor analysis exhibit certain regional characteristics.

Using a risk detector, the appropriate range of NDVI averages for each driving factor was determined (

Table A2. Suitable driving factor ranges or types with the highest average NDVI in Yunnan Province and its 16 cities.

Factors	Aspect	Elevation	Slope	Soil Type	Landform Type	Precipitation	Temperature	Nighttime Light	Population Density	Land Use Type
YN	North	(809.6, 1522.2]	(17.4, 32.0]	Leached soils	High-relief mountains	(1398.9, 2011.1]	(19.9, 24.6]	[0, 0.044]	[0, 4.0]	Forest
BS	North	(2338, 2719]	(28.0, 49.8]	Leached soils	High-relief mountains	(1469.0, 1729.8]	(11.8, 13.9]	(0, 0.042]	(0.3, 4.1]	Forest
CX	North	(2683, 3628]	(22.2, 48.8]	Leached soils	High-relief mountains	(890.5, 1093.4]	(6.8, 12.0]	(0, 0.044]	(6.1, 8.8]	Forest
DL	North	(2563, 2843]	(29.0, 51.9]	Leached soils	High-relief mountains	(1001.6, 1098.9]	(11.2, 13.0]	(0, 0.044]	(0, 3.5]	Forest
DH	Northeast	(1731, 1972]	(21.5, 40.8]	Ferralsols	High-relief mountains	(1762.3, 1966.2]	(17.2, 18.4]	(0, 0.034]	(0, 3.3]	Forest
DQ	North	(3462.8, 3809.6]	(28.3,56.7]	Leached soils	High-relief mountains	(769.1, 844.2]	(2.2, 4.8]	(0, 0.044]	(0, 0.52]	Forest
HH	North	(611.0, 963.0]	(29.1, 52.0]	Leached soils	High-relief mountains	(1491.8, 1842.6]	(19.9, 22.2]	(0, 0.044]	(0, 3.2]	Forest
KM	North	(2189.3, 2382]	(8.8, 16.3]	Leached soils	Moderate-relief mountains	(891.5, 929.1]	(13.5, 14.6]	(0, 0.045]	(0, 4.0]	Forest
LC	North	(2420.0, 3436.0]	(27.6, 49.1]	Alpine Soils	High-relief mountains	(1368.0, 1445.6]	(22.8, 24.0]	(0, 0.044]	(0, 7.0]	Forest
LJ	Flat	(2616, 3145]	(28.1, 56.2]	Leached soils	High-relief mountains	(786.9, 814.6]	(11.4, 14.9]	(0, 0.044]	(0, 1.2]	Forest
NJ	North	(1766.5, 2266.3]	(38.1, 53.6]	Ferralsols	High-relief mountains	(1227.8, 1328.1]	(8.3, 13.6]	(0, 0.038]	(2.8, 13.2]	Forest
PE	North	(2309, 3315]	(27.4, 48.7]	Alpine Soils	High-relief mountains	(1214.6, 1299.4]	(9.4, 14.7]	(0, 0.044]	(0, 3.8]	Forest
QJ	North	(754, 1812]	(9.1, 17.1]	Semi-Luvisols	Moderate-relief mountains	(1104.7, 1190.7)	(15.9, 21.8]	(0, 0.429]	(0, 3.3]	Forest
WS	West	(134, 940]	(26.6, 46.9]	Leached soils	Extreme-relief mountains	(1471.0, 1972.0]	(19.5, 23.3]	(0, 0.044]	(0, 3.1]	Forest
XSBN	North	(945.1, 1098.8]	(25.1, 43.7]	Leached soils	High-relief mountains	(1572.0, 1608.5]	(21.7, 22.4]	(0, 0.044]	(0, 2.9]	Forest
YX	North	(2407, 3047]	(27.2, 48.2]	Leached soils	Extreme-relief mountains	(1024.2, 1081.9]	(9.6, 11.7]	(0, 0.044]	(0, 3.5]	Forest
ZT	North	(1237.8, 1824.1]	(17.4, 32.0]	Semi-Luvisols	Moderate-relief mountains	(936.7, 990.2]	(14.5, 15.9]	(0, 0.044]	(0.1, 4.1]	Forest

). For the entire Yunnan Province, the highest annual average NDVI was observed when the aspect was north-facing, elevation ranged between 809.6 and 1522.2 m, slope was between 17.4 and 32.0°, soil type was leached soils, landform type was high-relief mountains, annual precipitation was between 1398.9 and 2011.1 mm, annual average temperature was between 19.9 and 24.6 °C, nighttime light was between 0 and 0.044, population density was between 0 and 4, and land use type was forest. Almost all cities in Yunnan Province have observed the highest NDVI under the north aspect and forest land use type. Aspect influences rainfall infiltration, runoff, and solar radiation absorption [112]. Shady and semi-shady slopes (north-facing slopes) have weaker evaporation and stronger water retention capabilities, which are conducive to moisture accumulation for vegetation growth. Therefore, compared to sunny and semi-sunny slopes (south-facing slopes), vegetation coverage is relatively higher [113]. Ten cities observed the highest annual average NDVI in leached soils. The high organic matter content of leached soils can continuously supply nutrients [114], and its stable structure and resistance to erosion together create a soil environment that is conducive to vegetation taking root and absorbing water and nutrients [113,115]. KM, QJ and ZT had higher NDVI in medium-relief mountains, while the remaining 13 cities had higher NDVI in higher-relief mountains. The steep terrain and low human industrial activity in these areas allow vegetation to grow naturally. Whether at the provincial or city scale, NDVI values are higher when the climate is warm and humid and population density is low. Under these conditions, vegetation experiences less human interference. As mentioned in previous studies, the growth of cities in central Yunnan has triggered increased encroachment on ecological spaces and resource environments, resulting in lower vegetation coverage compared to other areas [116].

To understand how explanatory variables jointly influence NDVI, we employed interaction detectors to evaluate if the synergy between any two drivers amplified or diminished their ability to account for NDVI patterns (Figure 6). Analysis across Yunnan revealed that synergistic effects occurred in all 45 pairs of factor interactions. Specifically, 15 pairs demonstrated a nonlinearly enhancement influence, and the remaining 30 exhibited enhancement from dual factors. This indicates that coupling factors substantially intensifies their role in shaping the spatial variation in NDVI. Within the province, the combined influence of rainfall and land cover proved most potent, accounting for over half (>50%) of NDVI’s spatial variance. Furthermore, land cover interactions with topographic position (aspect, elevation, slope), soil characteristics, geomorphology, climatic variables (rainfall, temperature), and socio-economic indicators (nighttime lights, population) each elucidated more than 40% of NDVI’s spatial pattern. These findings highlight land cover as the primary factor governing the spatial arrangement of NDVI in Yunnan. It is also noteworthy that the intensity of these synergistic effects differed markedly between urban areas.

Ecological detection demonstrated significant variation in the joint impact of any two explanatory variables on Yunnan’s NDVI changes (Figure 6). The ecological detector showed a statistically significant difference (p < 0.05) in the influence of any single factor compared to others, indicating they are closely interdependent and non-substitutable drivers [84]. Climate, topography, and socio-economic conditions collectively influence the spatial pattern and change trends of NDVI in Yunnan Province.

3.3. Time Lag Effect of Meteorological Indicators on NDVI Changes

Applying partial correlation analysis, we determined the maximum partial correlation coefficients (MPC) between NDVI and precipitation, as well as temperature across the study area (Figure 7). Grids lacking statistically significant correlations (p > 0.05) were excluded from both Figure 7 and subsequent analysis. As shown in Figure 7, the data reveal a distinct lagged pattern in how vegetation growth responds to climate drivers. Regarding precipitation, approximately 15% of the study region exhibited a dominant negative MPC with NDVI. These regions are concentrated in the northwestern cities (including DQ, NJ), the southwestern cities (including DH, BS, LC, PE and XSBN) and the southeastern cities (such as HH). Given the high precipitation levels in these zones, the significantly negative MPC between NDVI and precipitation may be related to factors that are unfavorable to vegetation growth, such as excessive soil moisture and root hypoxia caused by over-precipitation. Additionally, the complex geological structure and fragile environmental conditions in northwestern Yunnan make the region prone to geological disasters due to excessive rainfall. Regions with a positive MPC account for about 81% of the study area. Among these, only 5% of the study area has an MPC greater than 0.6, while nearly half (47%) of the study area is in the range of 0.3–0.6. Compared to other western cities, northeastern cities in Yunnan receive less rainfall, and most regions exhibit a positive MPC, indicating that precipitation substantially enhances vegetation development in these zones. Across 83% of Yunnan, the vegetation response to precipitation lags by 1–3 months (Figure 7b), averaging approximately 1.5 months. In the largest proportion of areas (36%), the vegetation response to precipitation lags by 1 month. In northwestern and southwestern cities such as DQ, NJ, and BS, the lag period is 0–1 months. Regions with a 3-month lag are mainly concentrated in southern Yunnan.

Regarding temperature, the MPC between NDVI and temperature is significantly positive in 74% of Yunnan, with these areas clustered within the province’s northeast and northwest sectors. These regions have lower temperatures compared to the southwest and provide thermally suitable environments that support optimal vegetation development. Additionally, Yunnan’s terrain gradually rises from the southeast to the northwest, and NDVI decreases with increasing altitude, mainly due to the temperature drop at higher elevations. Therefore, a moderate temperature increase in high-altitude areas is more beneficial to NDVI [117]. Moreover, the recent rise in nighttime temperatures in Yunnan has enhanced the resistance of vegetation communities to frost damage, positively impacting alpine vegetation [34]. Regions with a significantly negative MPC account for about 20% of the study area (mostly between −0.3 and 0), mainly distributed in the southwestern and central parts of Yunnan. Characterized by a tropical and subtropical monsoon climate, the southwest maintains high temperatures all year round, where excessively high temperatures may hinder vegetation growth or even cause vegetation death. Additionally, the central and western regions experience frequent droughts, especially after 2000, with drought events mainly occurring in central and western Yunnan [118]. Rising temperatures and associated drought events have adverse effects on vegetation. In 82% of Yunnan, the vegetation response to temperature was lagged by 1–3 months (Figure 7d), averaging 2.0 months. In some parts of northwestern, southwestern and eastern Yunnan, the lag period for vegetation response to temperature is 3 months.

3.4. The Contribution of Climate Change and Human Activities to NDVI Changes

From 2001 to 2023, synergistic impacts from climatic shifts and anthropogenic factors shaped NDVI’s spatial and temporal dynamics across Yunnan. According to Figure 8a, approximately 73% of the area experienced increasing NDVI trends is primarily attributable to these combined influences. Additionally, about 7% of the regions exhibited a combined effect leading to a decrease in NDVI, primarily distributed in areas such as BS, DL, CX, KM, YX, QJ, and HH, covering the central part of Yunnan Province. Areas where NDVI increases (decreases) resulted exclusively from climatic influences or anthropogenic forces constituted minor proportions: 2% (1%) and 13% (4%) respectively. The dominant beneficial role of human interventions on vegetation dynamics was principally clustered in the western and central parts of Yunnan (including cities such as DH, BS, YX, and HH).

Residual-based assessment quantified the relative influence of distinct drivers on NDVI patterns. As shown in Figure 8b, climate-induced NDVI increase accounts for about 92% in Yunnan Province. Among these, regions with climate change contribution rates in the ranges of 0–20% and 20–40% covered larger areas, accounting for approximately 53% and 25% of the study area, respectively. In ZT City, the contribution rate of climate change to NDVI changes exceeded 60%, and in some areas, it reached 80%. As illustrated in Figure 8c, human activities promoted NDVI increases in about 90% of the region. Notably, in some regions, the contribution rate of human activities to NDVI even exceeded 80%, highlighting their primary role in vegetation recovery. Conversely, in the central Yunnan region, where urbanization is highly intensive, the inhibitory effect of human activities on NDVI was mainly concentrated, with an average contribution rate generally below −20%. Overall, from 2000 to 2023, human activities drove positive vegetation succession in Yunnan Province. Through a series of ecological projects, China has significantly enhanced its vegetation coverage [119].

4. Discussions

4.1. Comparison with Previous Studies

Our analysis reveals a 0.025/decade upward trend in Yunnan’s annual max NDVI during 2001–2023. The multi-year average NDVI shows a spatial decreasing trend from southwest to southeast. The southwest region is characterized by low mountains, abundant precipitation and high vegetation cover. Approximately 84% of Yunnan Province exhibited an increasing trend in NDVI, while only 3% and 6% of the areas showed significant and insignificant decreases, respectively. These degraded areas were mainly concentrated in the central part of Yunnan Province, where rapid urbanization has exerted pressure on the ecological environment, such as overdevelopment and land cover changes, leading to a negative average trend in the proportional nonlinear contribution (PNC) to NDVI [120]. Overall, Yunnan’s vegetation is characterized by “improvement as the main trend with localized degradation,” consistent with previous studies [121,122]. Unlike previous studies, this study conducted a comprehensive assessment of all 16 prefecture-level cities in Yunnan Province. We found that Xishuangbanna had the highest multi-year average NDVI but a relatively slow vegetation growth rate, while DQ had the lowest multi-year average NDVI and growth rate. Nevertheless, all 16 cities showed varying degrees of NDVI increase. These results reflect the vegetation growth conditions of each prefecture-level city at a finer scale, offering a scientific basis for local governments to formulate targeted ecological protection policies.

Vegetation changes, local climate, and human disturbances exhibit complex nonlinear interactions [123]. This study evaluated the impacts of climate, topography, and human activities on NDVI in Yunnan Province using a geographical detector. Unlike previous studies that primarily focused on macro-level analyses of the entire province [122], this study, while maintaining an understanding of the overall NDVI trends and characteristics in Yunnan, further refined and supplemented the analysis by identifying the main driving forces of NDVI changes in each of the 16 prefecture-level cities. The results indicate that due to differences in natural environments and socio-economic development levels among cities, the factor analysis results exhibit regional specificity. Overall, land use type is the primary driver of NDVI spatial heterogeneity, and its interaction with precipitation significantly enhances this effect. However, in high-altitude areas, there is a strong correlation between terrain and NDVI. In the central Yunnan region, which is highly urbanized, NDVI spatial heterogeneity is more closely related to changes in population and nighttime light. Additionally, we identified the optimal ranges or types of driving factors for the highest average NDVI in Yunnan Province and its 16 cities. The results provide scientific support for local governments to formulate ecological protection and management measures.

Yunnan Province has an intricate and variable climate. Climate not only exerts complex nonlinear effects on vegetation changes but also exhibits a time-lag effect in vegetation’s response to climate. Studying this time-lag effect helps establish connections between atmospheric conditions and plant growth [124]. Our investigation identifies climate-response lags across 82% of Yunnan’s ecosystems, with average lag periods of 1.5 months for precipitation and 2.0 months for temperature. Vegetation in Yunnan is more sensitive to precipitation. Crucially, the meteorological influences primarily benefited the regional flora, which is consistent with the observations of Han et al. [125]. Residual analysis further disentangled anthropogenic and climate forcing mechanisms, revealing synergistic drivers behind the NDVI increase in 73% of the region. Recent conservation measures in China have significantly increased vegetation cover [119]. Thus, human activities primarily facilitate vegetation growth in Yunnan.

Aligned with vegetation dynamics in Yunnan Province, other subtropical mountain regions—including central Mexico and Fujian Province, China—have similarly exhibited increasing vegetation indices and regional greening trends in recent decades [126,127]. This convergence stems from shared drivers of favorable hydrothermal conditions and ecological restoration policies, revealing significant commonalities in both vegetation trajectories and underlying mechanisms. As a humid zone, our findings demonstrate that meteorological factors predominantly exert positive rather than inhibitory effects on Yunnan’s vegetation. In stark contrast, temperate arid/semi-arid regions and tropical dry ecosystems experience dominant negative accumulated temperature effects at 3-month lags. These areas exhibit high sensitivity to climate fluctuations coupled with limited resilience capacity [128], forming a distinct dichotomy with Yunnan. Concurrently, this study confirms land use type as the primary driver of NDVI spatial heterogeneity across Yunnan. Existing research substantiates that urbanization and land use conversions significantly impact forest productivity across all climatic zones [129], underscoring the predominant influence of land use modifications on vegetation dynamics.

4.2. Reflection on Regional Ecological Protection and Management Strategies

Changes in vegetation coverage levels are mainly associated with land use type changes [130], and land use transitions are the primary cause of NDVI loss and gain [131]. Forest soils have the highest organic matter content, benefiting soil productivity, water retention, and carbon sequestration [132]. Moreover, among different vegetation types, forests and shrubs exhibit stronger drought resistance and shorter recovery times after droughts compared to crops and grasslands [133]. As ecological barrier zones and biodiversity hotspots in northwestern and southwestern Yunnan, these regions exhibit high forest coverage. However, extensive planting of commercial tree species has posed severe threats to tropical forests and protected areas in recent years [41]. Consequently, forest conservation strategies must designate core protected zones and transboundary water conservation corridors as off-limits to development, while progressively restoring vegetation-depleted lands (e.g., grasslands, farmlands) to high-density forest cover [96]. In the southwest, optimizing ecological corridor networks for biodiversity refugia is essential to achieve synergistic conservation of ecosystems and species diversity [134,135]. Distinct hydro-climatic zones with corresponding vegetation habitats typically form along elevational gradients [136]. The high-altitude northwest requires prioritized climate adaptation strategies. Implementing vertical-gradient-optimized approaches—selecting tree species with high climatic adaptability and carbon sequestration potential—will enhance regional forest resilience [137].

Central and eastern Yunnan, characterized by intensive human activities, necessitate balanced ecological governance integrating water resource management, land use planning, and food security. Metropolitan development must be reconciled with ecosystem integrity, pursuing both economic advancement and environmental enhancement [96]. Urban greening strategies should select climate-adapted species and optimize spatial configurations to strengthen ecological quality [138]. Coordinated urban–rural resilience planning with regionally tailored pathways will improve risk resistance capacity [139], preventing environmental justice issues arising from green space disparities in low-income communities. The karst landform area in eastern Yunnan is relatively large, and some studies have shown that ecological restoration projects (ERPs) in the karst regions of Yunnan have been less effective, likely due to low rainfall, poor management practices (e.g., inappropriate species selection, low farmer compensation rates), and other factors [140]. Moreover, as global warming intensifies, the effectiveness of current ecological restoration management practices may decline, potentially slowing or halting greening trends [117]. Therefore, in the future, ecological restoration projects need to adopt more scientific, rational, and adaptive strategies to address these challenges, taking into account natural conditions, management practices, and climate change to promote high-quality development in Yunnan Province.

4.3. Limitations and Uncertainties

This study reveals and quantifies the spatiotemporal patterns of NDVI across Yunnan Province, disentangling the effects of climatic, topographic, and anthropogenic drivers. However, several limitations warrant acknowledgment. First, while the adopted NDVI product underwent rigorous Savitzky–Golay filtering and noise reconstruction to ensure reliability, NDVI inherently fails to capture vegetation-type transitions, community structure dynamics, or retrogressive succession in grasslands [141]. Future investigations should integrate complementary vegetation indices to characterize ecological changes while mitigating NDVI saturation effects in herbaceous ecosystems [142]. In data-scarce mountainous regions (e.g., northwestern Yunnan), uncertainties arise from limited meteorological, hydrological, and land use data availability [48]. The critical shortage of high-elevation weather stations with precipitation monitoring capabilities remains an obstacle for accurate mountain climate quantification [24]. Finally, future research should develop integrated modeling frameworks to systematically investigate the coupled relationships among NDVI, climate variability, and human disturbances.

5. Conclusions

In this study, we focused on Yunnan Province, China, and investigated the spatiotemporal dynamics of NDVI from 2001 to 2023 at both provincial and municipal scales using MK trend analysis. We also explored the main driving factors of NDVI changes through geographic detector, time-lag analysis, and residual analysis. The main conclusions are as follows:

(1)

From 2001 to 2023, the multi-year average NDVI in Yunnan Province was 0.80, with 98% of the regions having an annual average maximum NDVI > 0.6. Spatially, NDVI decreased from the southwest to the southeast. The annual maximum NDVI for the entire province increased at a rate of 0.025 per decade. Meanwhile, the annual maximum NDVI in all 16 prefecture-level cities showed an increasing trend, with QJ city experiencing the fastest growth rate (0.036 per decade) and DQ city the slowest (0.012 per decade). The decrease in NDVI was primarily concentrated in the central cities of Yunnan.

(2)

In multi-spatial scale analysis, land use type is the primary determinant of NDVI spatial heterogeneity. The interaction between land use type and precipitation was the strongest, explaining over 50% of the spatial distribution of NDVI. For the entire Yunnan Province, the highest annual average NDVI was observed when the aspect was north-facing, elevation ranged between 809.6 and 1522.2 m, slope was between 17.4 and 32.0°, soil type was leached soils, landform type was high-relief mountains, annual precipitation was between 1398.9 and 2011.1 mm, annual average temperature was between 19.9 and 24.6 °C, nighttime light was between 0 and 0.044, population density was between 0 and 4, and land use type was forest.

(3)

Precipitation showed significant positive associations with NDVI across 81% of Yunnan, which promoted plant development in northeastern zones. In contrast, 15% of the region showed a negative precipitation–NDVI relationship, concentrated in the northwest, southwest and southeast. In 83% of the regions, vegetation responded to precipitation with a lag of 1–3 months, averaging 1.5 months. Temperature conditions showed a positive correlation with NDVI in 74% of the province. A negative temperature–NDVI relationship was observed in 20% of the region. The temperature response delay lasted 1–3 months (mean 2.0 month) in 82% of the region. This analysis revealed different spatiotemporal patterns in climate–vegetation interactions.

(4)

Across the province, synergistic climate–anthropogenic influences drove NDVI increases in 73% of study area and decreases in 7%. Enhancement effects from climatic forces and human activities drivers dominated approximately 92% and 90% of vegetated areas respectively. From 2000 to 2023, human activities primarily promoted vegetation growth in Yunnan, with urbanization in the central region being more intense than in other areas. The inhibitory effects of human activities on NDVI were mainly concentrated in this region.

(5)

As critical ecological security barriers and biodiversity hotspots, northwestern and southwestern Yunnan should establish forest conservation strategies with designated protected areas, optimize ecological corridor networks to create biodiversity refugia, and implement elevation-specific climate adaptation measures. The central and eastern regions—characterized by intensive human activities—should develop urban greening initiatives to enhance coordinated urban–rural ecological resilience, while dynamically adjusting critical phases of ecological restoration projects.