Assessing the Distribution and Driving Effects of Net Primary Productivity along an Elevation Gradient in Subtropical Regions of China

Abstract

Globally, forest ecosystems, especially subtropical forests, play a central role in biogeochemical cycles and climate regulation, demonstrating their irreplaceable function. The subtropical region of China, characterized by its unique forest ecosystem, complex terrain, climate heterogeneity, diverse vegetation types, and frequent human activities, underscores the importance of the in-depth study of its net primary productivity (NPP). This paper employs the eddy covariance–light use efficiency (EC-LUE) model to quantitatively estimate the gross primary productivity (GPP) of this region from 2001 to 2018, followed by an estimation of the actual net primary productivity (ANPP) using the carbon use efficiency (CUE). The results showed that over these 18 years, the annual average ANPP was 677.17 gC m⁻² a⁻¹, exhibiting an overall increasing trend, particularly in mountainous areas, reserves, and the cultivated lands of the northeastern plains, whereas a significant decrease was observed around the urban agglomerations on the southeast coast. Furthermore, the Thornthwaite memorial model was applied to calculate the potential net primary productivity (PNPP), and diverse scenarios were set to quantitatively evaluate the impact of climate change and human activities on the vegetation productivity in the study area. It was found that in areas where the ANPP increased, both human activities and climate change jointly influenced ANPP dynamics; in areas with a decreased ANPP, the impact of human activities was particularly significant. Additionally, the heterogeneous distribution of ANPP across different altitudinal gradients and the driving effects of various climatic factors were analyzed. Finally, a partial correlation analysis was used to examine the relationships between the temperature, precipitation, and ANPP. This study indicated that temperature and precipitation have a substantial impact on the growth and distribution of vegetation in the region, yet the extent of this influence shows considerable variation among different areas. This provides a robust scientific basis for further research and understanding of the carbon dynamics of subtropical forest ecosystems and their role in the global carbon cycle.

1. Introduction

Forests are an extremely important component of terrestrial ecosystems [1], playing a vital role in influencing the global carbon cycle. They store nearly two-thirds of terrestrial carbon and have a carbon density higher than other land use types [2]. As the largest carbon sink in terrestrial ecosystems, their contribution to maintaining the global carbon balance, reducing the atmospheric accumulation of greenhouse gases such as CO₂, and preserving the stability of the global climate is unparalleled [3,4]. Furthermore, subtropical forest ecosystems occupy a significant position in the global forest ecosystem, constituting a critical part of the forest carbon pool. They play a very crucial role in global terrestrial ecosystems [5], are intricately linked with global biogeochemical cycles [6], and exert a certain regulatory influence on the climate [7,8]. Subtropical forest ecosystems encompass a crucial segment of global vegetation diversity and are among the most productive ecosystems on Earth [3,9].

On a global scale, climate warming and the rising concentration of carbon dioxide are among the primary characteristics of climate change. These changes have triggered a series of environmental issues, including ecosystem degradation, an increase in the frequency of extreme weather events, and land desertification. These environmental challenges typically occur concurrently with global warming, and their synergistic effects profoundly impact forest growth, mortality, and succession processes, thereby severely disrupting the carbon sequestration capabilities and other ecological functions of forests [10,11,12]. Concurrently, escalating human activities, such as uncontrolled resource extraction, human-induced forest fires, the invasion of alien species, and forest clearance due to population expansion, exert considerable adverse effects on forest ecosystems [13,14]. Research on the global carbon cycle, especially in the context of climate change and human activities, has become a core issue in the field of environmental science [15,16,17,18,19,20,21].

Vegetation is a vital component of forests, playing a crucial role in the energy and material exchange processes within ecosystems [22]. The productivity of vegetation can be segregated into gross primary productivity (GPP) and net primary productivity (NPP), being conceptual frameworks developed in the course of scholars investigating plant productivity. GPP refers to the rate at which green plants in the ecosystem synthesize organic substances by assimilating carbon dioxide through photosynthesis and absorbing solar energy [23]. NPP represents the accumulation of organic matter per unit area and unit time, characterized as the portion of organic carbon fixed through photosynthesis, minus the part consumed through plant respiration [24]. Vegetation productivity reflects the process where plants absorb CO₂ from the atmosphere through photosynthesis, converting light energy to chemical energy, while accumulating organic dry matter. It embodies the production capability of terrestrial ecosystems under natural conditions, serving as an important ecological indicator for estimating Earth’s carrying capacity and evaluating the sustainable development of ecosystems. It is also a primary metric for determining the carbon sink of ecosystems and regulating ecological processes [25].

Currently, there are numerous methods to estimate the primary productivity of forest vegetation, and the light use efficiency (LUE) model is the main method for estimating vegetation productivity based on remote sensing data. This principle was first proposed by Lieth while studying crop productivity. The LUE model has clear principles and simple calculation processes. The input data for the model can mostly be acquired through remote sensing and meteorological stations, and it offers high simulation accuracy, becoming a primary tool for regional and global vegetation productivity assessments [26]. The LUE model mainly encompasses methods such as the Carnegie–Ames–Stanford approach (CASA) model, the eddy covariance–light use efficiency (EC-LUE) model, the GLO-PEM model, and the photosynthesis model (VPM), among others. Zhang et al. utilized the CASA model to simulate the net primary productivity of vegetation in the Three River Source region of China from 1982 to 2012, and analyzed the main climatic and human driving factors [27]. Xu et al. referred to the drought-stress improved-EC-LUE model to estimate the spatiotemporal distribution pattern of GPP in the Mao bamboo forest of Anji City [28]. Martínez et al. applied the GLO-PEM model to estimate the net primary productivity of the Iberian Peninsula for 15 years and studied its spatial and temporal variations and the quantitative relationship between NPP, precipitation, and air temperature [21]. Although many scientific studies have used light use efficiency models to study vegetation productivity, currently most LUE models directly simulate the GPP, with only the CASA and GLO-PEM models capable of directly simulating the NPP [23]. Theoretically, the fundamental principle of the LUE model is designed for GPP; the direct simulation of NPP requires the assumption that the proportion of plant autotrophic respiration to GPP is the same across all ecosystem types and geographic regions. However, research indicates that the proportion of plant autotrophic respiration to GPP is not the same under temperature gradients, and using the same ratio for calculation will inevitably lead to corresponding errors [23]. Therefore, this study uses the EC-LUE model to estimate the total primary productivity in the research area, and uses the ratio of NPP to GPP as the value of carbon use efficiency (CUE), utilizing the product of GPP and CUE to represent the net primary productivity of vegetation in the research area.

2. Materials and Methods

2.1. Study Area

The subtropical region of China is situated between 21° and 35° north latitude, extending from 22°30′ to 21°30′ north in the south, and nearing 35° north in the north, aligning with Qinling Mountain, Huai River, and Bailong River [29]. This vast region traverses the central part of Taiwan Island and the southern part of the Leizhou Peninsula, encompassing the area south of the Qinling Mountain–Huaihe River line and most of the region east of the Hengduan Mountains, including 20 provinces (Figure 1). The subtropical zone in China is a unique forest ecosystem, characterized by complex topographical heterogeneity and climate features: a topography that is higher in the east and lower in the west, with complex landform types, situated between China’s second and third terraces, mostly belonging to the lowest terrace (Figure 1) [30]. This region enjoys superior hydrothermal conditions, fostering a rich diversity of forest types and species diversity [6]. Overall, evergreen broadleaf forests predominantly occupy the subtropical regions of China [31].

2.2. Data Collection and Preprocessing

2.2.1. Data to Calculate GPP, NPP, and PNPP

The photosynthetically active radiation (PAR) data used in this study came from the Global Land Surface Feature Parameters (GLASS) data product, developed independently by the team of Professor Liang at Beijing Normal University. The GLASS PAR data product is derived from the inversion of Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR) data, wherein the AVHRR data are estimated using an improved lookup table algorithm and MODIS data are estimated using a hybrid algorithm, providing daily data with a resolution of 0.05° [32,33,34].

The Normalized Difference Vegetation Index (NDVI) is a commonly used vegetation index and is one of the essential parameters reflecting the growth and nutritional information of crops; it is extensively used in estimating forest productivity [35]. MODIS Normalized Difference Vegetation Index (NDVI, MOD13A1 V6.1) is a continuous index of NDVI derived from NOAA-AVHRR, calculated based on the bidirectional surface reflectance that has undergone atmospheric correction. These reflectances have excluded water, clouds, heavy aerosols, and cloud shadows, according to the dataset which is of high quality and can objectively reflect vegetation dynamics. The NDVI dataset originates from Google Earth Engine (http://earthengine.google.com, accessed on 25 July 2023), with a time resolution of 16 days and a spatial resolution of 500 m, built using the Maximum Value Composite (MVC) method to construct monthly NDVI data [36], further eliminating errors caused by clouds, atmosphere, solar angles, and other interfering factors [37].

The average temperature and dew point temperature data were both sourced from ERA-LAND, a reanalysis dataset providing a consistent view of the evolution of land variables over several decades, at a higher resolution compared to ERA5. The temperature and dew point temperature data provided by ERA are in Kelvin units, which can be converted to Celsius by subtracting 273.15 during specific calculations; the temporal and spatial resolutions are monthly and 0.1°, respectively.

The precipitation data utilize version 4 of the precipitation data provided by the Climatic Research Unit (CRU), which was cross-verified at the station level to ensure its accuracy; the temporal and spatial resolutions are monthly and 0.5°, respectively. The annual total precipitation was calculated from the accumulation of monthly data. The Digital Elevation Model (DEM) data were acquired from the SRTM DEM dataset provided by the National Qinghai-Tibet Plateau Data Center, with the dataset primarily being the fourth version created by CIAT (International Center for Tropical Agriculture), utilizing a new interpolation algorithm [38] which better fills the data gaps of SRTM 90. DEM refers to the Digital Elevation Model, which is a physical representation of the terrain’s elevation, depicted through an array of ordered numerical values. This method has more effectively filled the data gaps of SRTM 90 (accessed on 26 May 2023 from https://data.tpdc.ac.cn/zh-hans/data/acb49ce8-2bfe-4ab4-97ff-e6e727110703).

The land cover dataset used was the MODIS Land Cover Type (MCD12Q1) version 6.1 data product, which offers global land cover types at an annual interval. The MCD12Q1 6.1 version data product is derived through the supervised classification of MODIS Terra and Aqua reflectance data, and this study used its International Geosphere-Biosphere Programme (IGBP) classification scheme, with temporal and spatial resolutions of annually and 500 m, respectively.

2.2.2. Data to Calculate Carbon Use Efficiency

The carbon use efficiency (CUE) of terrestrial ecosystems is defined as the ratio of net primary production (NPP) to gross primary production (GPP) [37,38], a critical metric for measuring carbon transformation, depicting the capability of ecosystems to transfer carbon from the atmosphere to terrestrial biomass [39].

This study utilized GPP data from the Global Land Surface Satellite (GLASS) data products. The GLASS-GPP algorithm originates from the EC-LUE model [40], incorporating the impacts of several significant environmental variables such as atmospheric carbon dioxide concentration, radiation components, and vapor pressure deficit (VPD), which can effectively replicate spatial, seasonal, and annual variations, especially enhancing the model’s ability to represent annual variations in GPP [41]. The data are available on an annual basis with a resolution of 500 m.

Meanwhile, the NPP data were derived from the MODIS NPP product (MOD17A3HGF) version 6.1, which offers NPP data with a 500 m resolution. The annual net primary production is calculated by summing all 8-day net photosynthesis (PSN) products (MOD17A2H) of a specified year. The PSN values are determined by the difference between the gross primary production (GPP) and maintenance respiration (MR), presented as annual data with a resolution of 500 m.

It is important to add that all data were processed to have a spatial resolution of 500 m, facilitating better calculations.

2.3. Methods

2.3.1. Estimation of GPP and ANPP

In this study, we utilized the eddy covariance–light use efficiency (EC-LUE) model to estimate the GPP of subtropical ecosystems in China. The EC-LUE model is based on two assumptions: firstly, the fraction of absorbed PAR (FPAR) is a linear function of NDVI; secondly, the actual light use efficiency calculated according to the constant potential LUE unrelated to the organism is controlled by temperature or soil moisture, taking the maximum of the two restrictions. The EC-LUE model was calibrated and validated using 24,349 daily GPP estimates from 28 eddy covariance flux towers in the United States and Europe, covering a range of forests, grasslands, and tropical savannas. The model explained 85% and 77% of the observed daily GPP variations at all calibration and validation sites, respectively. Compared with the GPP calculated using the Moderate Resolution Imaging Spectroradiometer (MODIS), the GPP predicted using the EC-LUE model was better matched with the tower data at these sites [39].

The EC-LUE model equation is as follows:

$$GPP={\epsilon }_{max}\times PAR\times FPAR\times min(T_s,W_s)$$

(1)

where ${\epsilon }_{max}$ is the maximum light utilization efficiency, which is 2.25 gC/MJ [40]; $PAR$ is the photosynthetically active radiation; $FPAR$ is the photosynthetically active radiation component, which is calculated using the linear function of NDVI [41]; and $T_s$ and $W_s$ are the limiting effects of temperature and water on the potential light utilization rate, respectively. The EC-LUE model holds that the limiting effect of the two factors follows the minimum factor rule of ecology, that is, the ultimate environmental limit depends on the environmental factor with the strongest stress.

In reference to existing research results, the calculation formulas for $FPAR$, $T_s,$ and $W_s$ are as follows:

$$FPAR=a\times NDVI+b$$

(2)

$$T_s=\frac{\left(T-T_{min}\right)\left(T-T_{max}\right)}{\left(T-T_{min}\right)\left(T-T_{max}\right)-{\left(T-T_{opt}\right)}^2}$$

(3)

$$W_s=\frac{VP{D}_0}{\left(VP{D}_0+VPD\right)}$$

(4)

where $a$ and $b$ are empirical coefficients, valued at 1.24 and −0.168, respectively. This formula and parameters have been verified in various ecosystems globally, showcasing extensive representativity [42]. $T$ stands for air temperature (°C), and $T_{min}$, $T_{max}$, and $T_{opt}$ represent the minimum, maximum, and optimum temperatures for plant photosynthesis, with values of 0 °C, 40 °C, and 21 °C, respectively [40]. $VP{D}_0$ is the empirical coefficient for the $VPD$ constraint equation, with each vegetation type having a corresponding $VP{D}_0$. The calculation formula for $VPD$ (vapor pressure deficit) is as follows:

$$VPD=SVP-AVP$$

(5)

$$SVP=6.112\times f_w\times e^{\frac{17.67T}{T+253.5}}$$

(6)

$$AVP=6.112\times f_w\times e^{\frac{17.67T_d}{T_d+253.5}}$$

(7)

where $SVP$ and $AVP$ are saturated vapor pressure and actual vapor pressure (kPa), respectively, $T$ is the air temperature (°C), and $T_d$ is the dew point temperature (°C).

$$f_w=1+7\times {10}^{-4}+3.46\times {10}^{-6}{P}_{mst}$$

(8)

$$P_{mst}=P_{msl}{\left(\frac{T+273.16}{T+273.16+0.0065\times Z}\right)}^{5.625}$$

(9)

where $Z$ stands for elevation (m), $p_{mst}$ is the atmospheric pressure (hPa), and $p_{mst}$ refers to the mean sea-level pressure, valued at 1013.25 hPa.

We calculated ANPP through carbon use efficiency (CUE), which is defined in terrestrial ecosystems as the ratio of net primary productivity (NPP) to gross primary productivity (GPP) [43,44]. Usually, the value of CUE is defined as a constant 0.5, but with the increase in research and further refinement, a large number of studies have shown that the value of CUE varies with environmental factors, forestland factors, and changes in the ecosystem [45,46]. Therefore, this article selected NPP and GPP as the calculation method for CUE, and after calculating the pixel-scale CUE index values, the value of NPP was obtained from the following formula:

$$ANPP=GPP\times CUE$$

(10)

2.3.2. Validation of GPP and ANPP

The EC-LUE model was refined to enhance the accuracy of GPP assessments, consequently improving the precision of ANPP estimations. In the EC-LUE model used for the GLASS GPP product, average temperature data were derived from Modern Era Retrospective Analysis for Research and Applications (MERRA) with a spatial resolution of 0.5° [40]. However, our study employed average temperature data from ERA-LAND, offering superior spatial resolution compared to MERRA’s temperature data.

However, as the actual field data cannot directly calculate the GPP values, it is not feasible to determine the accuracy of GPP directly from field data. Therefore, we first compared the calculated GPP values with the product GPP values to ensure the rationality of the calculated GPP. Subsequently, we converted the biomass observed in the field plots to obtain the NPP of the sampled area. This ANPP was then compared with both the calculated ANPP and the product ANPP to evaluate the accuracy of ANPP, thereby ascertaining the accuracy of GPP. The field-derived NPP used to verify and enhance the precision of the EC-LUE model was sourced from field survey data within the study area. Twelve sites were surveyed, with four 1 m × 1 m plots established at each site. The above-ground parts were harvested and dried in a constant temperature oven until a constant dry weight was obtained, following which the carbon content was measured.

By comparing the calculated GPP with the product GPP, an R² value of 0.854 was achieved (see Figure 2), demonstrating the reasonableness of the calculated GPP for estimating ANPP. Next, the ANPP derived from GPP calculations was compared with the NPP obtained from field sites, yielding an R² of 0.791 (Figure 3a). In contrast, comparing the product NPP with field-derived NPP results gave an R² of only 0.768 (Figure 3b). Therefore, it can be inferred that the optimized EC-LUE model provides a more accurate simulation of ANPP than the product NPP.

2.3.3. Estimates of PNPP and HNPP

The Thornthwaite memorial model, widely used for estimating PNPP [47], calculates PNPP based on the annual actual evapotranspiration [48,49,50]. This model, which takes temperature and precipitation data as inputs, is an improved version of the Miami model and is known for its accuracy [51]. It is calculated using the following formula:

$$PNPP=3000\times \left[1-e^{-0.0009695\times \left(v-20\right)}\right]$$

(11)

$$v=\frac{1.05\times r}{\sqrt{1+{\left(1+\frac{1.05\times r}{L}\right)}^2}}$$

(12)

$$L=3000+25\times t+0.05\times t^3$$

(13)

where $v$ represents the annual actual evapotranspiration (mm), $L$ is the annual average evapotranspiration (mm), and $r$ and $t$ are the annual total precipitation (mm) and annual average temperature (°C), respectively.

HNPP reflects the impact of human activities on vegetation productivity [52,53] and is estimated by calculating the difference between PNPP and ANPP. The calculation formula for HNPP is as follows:

$$HNPP=PNPP-ANPP$$

(14)

2.3.4. Theil–Sen Slope Analysis and Mann–Kendall Trend Test

The Theil–Sen Median method is a robust non-parametric statistical trend calculation method. This method has high computational efficiency and is not sensitive to measurement errors and outlier data, making it suitable for long-term time-series data trend analysis, and it has been widely used in scientific research [54,55,56]. In this article, a Theil–Sen median estimation trend analysis was combined with a Mann–Kendall test for time-series analysis.

(1)

Theil-Sen median estimator

$$\beta =Median\left(\frac{x_j-x_i}{j-i}\right),(1\le i<j\le k)$$

(15)

where $\beta$ denotes the slope of changes involved in the time series under calculation, with k = 18 representing the length of the time series. $J$ and $i$ represent the years. When $\beta$ > 0, it indicates an upward trend in the time series; when $\beta$ < 0, it suggests a downward trend in the series.

(2)

Mann–Kendall trend test:

$$S={{\displaystyle \sum }}_{i=1}^{n-1}{{\displaystyle \sum }}_{j=i+1}^{n}sgn\left(x_j-x_i\right)$$

(16)

$$sgn\left(x_j-x_i\right)=\left\{\begin{array}{c}+1 x_j-x_i>0\\ 0 x_j-x_i=0\\ -1 x_j-x_i<0\end{array}\right.$$

(17)

$$Z=\left\{\begin{array}{c}\frac{S}{\sqrt{Var\left(S\right)}} (S>0)\\ 0 \left(S=0\right)\\ \frac{S}{\sqrt{Var\left(S\right)}} (S<0)\end{array}\right.$$

(18)

$$Var\left(S\right)=\frac{n\left(n-1\right)\left(2n+5\right)-{{\displaystyle \sum }}_{m=1}^n{t}_m\left(m-1\right)\left(2m+5\right)}{18}$$

(19)

where $x_j$ and $x_i$ represent the observed values corresponding to the time series $i$ and $j$ ($i$ < $j$), $sgn()$ is the sign function, and $Z$ signifies the standardized test statistic. $n$ is the number of entities in the time series ($n$ > 10), and $t_m$ indicates the range for any given mean m.

When the absolute value of $Z$ is greater than or equal to 1.96 (or 2.32), it denotes that the trend in the time series has been confirmed at a confidence level of 95% (or 99%). The experimental results passed a two-tailed significance test with a 95% confidence interval (confidence level, α = 0.05).

2.3.5. Scenario Establishment

The slope values of ANPP, PNPP, and HNPP represent the trend of changes at the pixel level during the research period. A positive slope of PNPP indicates that climate change has a positive impact on ANPP, whereas a negative value signifies its negative effect on ANPP. A positive slope for HNPP signifies that human activities have reduced ANPP, while a negative value indicates an increase in ANPP due to human activities. Based on the slopes of ANPP, PNPP, and HNPP, six scenarios have been identified to quantify the effects of climate change and human activities on ANPP (

Table 1. Eight scenarios of climate change and human activity impacts on ANPPs.

SANPP	SPNPP	SHNPP	Scenario	Diving Force	Abbreviation
>0	>0	>0	1	Climate	CDI
	<0	<0	2	Human	HDI
	>0	<0	3	Climate and Human	CHI
	<0	>0	4	Error	ER
<0	<0	<0	5	Climate	CDD
	>0	>0	6	Human	HDD
	<0	>0	7	Climate and Human	CHD
	>0	<0	8	Error	ER

). These include increases or decreases in ANPP due to climate change (CDI, CDD), human activities (HDI, HDD), as well as the combined influence of climate change and human activities (CHI, CHD).

2.3.6. Partial Correlation Analysis

To better understand the response mechanisms of ANPP, annual average temperature, and total precipitation, we calculated the partial correlation coefficients at the pixel scale to analyze the situation when the annual average temperature and total precipitation are correlated with ANPP at the same time. By eliminating the influence of one indicator, we calculated the correlation between the other two variables [57]. The calculation formula is as follows:

$$R_{\left(i,j|h\right)}=\frac{R_{ij}-R_{ih}\times R_{jh}}{\sqrt{\left(1-R_{ih}^2\right)\left(1-R_{jh}^2\right)}}$$

(20)

where $R_{ij}$ is the correlation coefficient between variables $x_j$ and $x_i$, Rih is the correlation coefficient between variables $x_i$ and $x_h$, $R_{jh}$ is the correlation coefficient between variables $x_j$ and $x_h$, and $R_{(\mathrm{i},\mathrm{j}|\mathrm{h})}$ is the partial correlation coefficient between $x_i$ and $x_j,$ excluding the effect of variable $x_h$.

3. Results

3.1. Spatiotemporal Variations in ANPP, PNPP, and HNPP

In this study, we conducted a temporal and spatial analysis of the computed ANPP, PNPP, and HNPP from 2001 to 2018, spanning 18 years.

Generally, the ANPP gradually increased from the northern to the southern part of the study area (Figure 4a). This trend became even more evident when we calculated regional statistics by province (Figure 4d). Meanwhile, the PNPP displayed a clear spatial distribution pattern (Figure 4b), incrementally increasing from the northern to the southern parts of the research area, displaying a near-circular distribution. The HNPP, on the other hand, featured a distribution characteristic of being lower in the west and higher in the east (Figure 4c).

From 2001 to 2018, the average ANPP in the research area was 677.17 gC m⁻² a⁻¹, showing an upward trend with an annual average increase of 4.84 gC m⁻² a⁻¹. The average PNPP was 1581.93 gC m⁻² a⁻¹, also showing an increasing trend, but with a lower annual growth rate of 0.54 gC m⁻² a⁻¹. The average HNPP was 900.44 gC m⁻² a⁻¹, displaying a decreasing trend with an annual average decline of −4.31 gC m⁻² a⁻¹ (Figure 5).

On a pixel scale, the proportions of different ANPP values were as follows: <250 gC m⁻² a⁻¹ accounted for 7.24%, 250–500 gC m⁻² a⁻¹ for 22.07%, 500–750 gC m⁻² a⁻¹ for 30.63%, 750–1000 gC m⁻² a⁻¹ for 25.42%, 1000–1200 gC m⁻² a⁻¹ for 9.40%, and >1200 gC m⁻² a⁻¹ for 5.24%. Based on the distribution across different ranges, it can be observed that most of the ANPP values were concentrated in the 250–1000 gC m⁻² a⁻¹ range, accounting for 76.12% of the total pixels.

The proportions of pixels for different PNPP values were as follows: <400 gC m⁻² a⁻¹ accounted for 3.82%, 400–800 gC m⁻² a⁻¹ for 1.08%, 800–1200 gC m⁻² a⁻¹ for 7.00%, 1200–1600 gC m⁻² a⁻¹ for 41.63%, 1600–1800 gC m⁻² a⁻¹ for 30.32%, and >1800 gC m⁻² a⁻¹ for 16.15%. The majority of PNPP values were primarily concentrated in the 1200–1800 gC m⁻² a⁻¹ range, making up 71.95% of the total pixels.

The distribution of pixel proportions across different HNPP value ranges was as follows: less than 0 gC m⁻² a⁻¹ constituted 0.88%, 0–300 gC m⁻² a⁻¹ constituted 7.61%, 300–600 gC m⁻² a⁻¹ constituted 16.36%, 600–900 gC m⁻² a⁻¹ constituted 35.33%, 900–1200 gC m⁻² a⁻¹ constituted 25.16%, 1200–1500 gC m⁻² a⁻¹ constituted 11.24%, and over 1500 gC m⁻² a⁻¹ constituted 3.42%; these values were predominantly concentrated in the 300–1500 gC m⁻² a⁻¹ range, accounting for 88.09% of the total pixels.

3.2. Analysis of Temporal and Spatial Trends in ANPP

From 2001 to 2018, 24.28% of the total area in the study region showed a significant increase in ANPP (Figure 6). In Yunnan province, the distribution was found in the southeast part, the western areas of Ailao and Wuliang mountains, the southern side of the northeast’s Wulian peak, and across both southern and northern sides of Wumeng mountain, which spans Yunnan and Guizhou provinces. In Sichuan province, the significant increase was mainly concentrated on the southeast side of the Daliang mountains in the southern area. Guizhou province exhibited a centralization on the eastern side of Wumeng mountain and within the southwest karst ecological protection area. Due to its proximity to Sichuan and Guizhou, Chongqing city showcased a distribution at the junction of these three provinces, notably in the northern Daliang mountain and the ecological protection area of the Three Gorges Reservoir in Chongqing. Shaanxi province was characterized by a significant distribution on the northern side of the central section of Daba mountains and in the southern part of Qinling mountainous ecological protection area. In Hubei province, the areas of notable increase encompassed Wushan and both the southern and northern flank of the eastern segment of Daliang mountain, the central part of the South-to-North Water Diversion Project source ecological conservation area, and the Dabie mountainous ecological protection zone, in addition to the northern region of Dahong mountain and the northern part of southern Mufu mountain. The neighboring provinces of Henan and Anhui demonstrated a dense distribution at the junction where these provinces meet with Hubei, primarily focused in the central region of Henan and the Huai River ecological protection area, with Anhui showcasing a concentration mainly on both sides of Zhangba Ridge and in the central and northern parts of the province. In Hunan province, which is situated in the central part of the study area, the main distribution areas were both sides of the Wuling mountains and the northern section of the eastern part of Xuefeng mountains. In contrast, only 4.89% of the area experienced a significant decrease (Figure 6), which was primarily found in the middle and lower Yangtze River basin, the Pearl River Delta urban cluster, and the central part of Taiwan.

3.3. Spatial Heterogeneity of ANPP at Altitude Gradient

The DEM of the research area was divided at intervals of 100 m, with each section constituting an elevation warehouse, to calculate the average ANPP and explore its spatial heterogeneity across elevation gradients. Overall, the ANPP of the research area demonstrated a significant increase, stabilizing with slight fluctuations before significantly decreasing as the elevation gradients changed. As shown in Figure 7, from 0 to 400 m, the ANPP presented a significant positive growth trend, increasing at a rate of 52.98 gC m⁻² a⁻¹ 100m⁻¹ (R² = 0.72, p < 0.05), growing from 526.02 gC m⁻² a⁻¹ to 737.95 gC m⁻² a⁻¹. Between 400 and 1600 m, the ANPP was distributed between 725.31 gC m⁻² a⁻¹ and 750.54 gC m⁻² a⁻¹, averaging 736.53 gC m⁻² a⁻¹, with a change of only 0.87 gC m⁻² a⁻¹, indicating that the ANPP was stabilizing. Afterwards, the ANPP enters a period of fluctuation, where from 1600 to 2800 m it initially decreases slightly between 1600 and 1900 m, dropping from 742.93 gC m⁻² a⁻¹ to 705.98 gC m⁻² a⁻¹. Then, between 1900 and 2800 m, it exhibited an upward trend, increasing at a rate of 7.71 gC m⁻² a⁻¹ per 100 m from 705.98 gC m⁻² a⁻¹ to 783.08 gC m⁻² a⁻¹, though the overall fluctuation during this period was not significant, with an average rate of change of 3.35 gC m⁻² a⁻¹ per 100 m. At elevations above 2800 m, the ANPP significantly decreased with the rise in elevation gradient (R² = 0.98, p < 0.001), rapidly dropping from the peak value of 783.07 gC m⁻² a⁻¹ to below 10 gC m⁻² a⁻¹. Beyond an elevation of 5000 m, the average value of ANPP was only 3.67 gC m⁻² a⁻¹. Within the 2800 to 5000 m range, the ANPP decreased at a rate of 32.41 gC m⁻² a⁻¹ per 100 m from 783.07 gC m⁻² a⁻¹ to 37.84 gC m⁻² a⁻¹.

3.4. Impacts of Climate Change and Human Activities on ANPP

Climate change and human activities have both influenced the ANPP of the research area. The evident spatial heterogeneity in ANPP at the pixel level indicates that climate change and human activities have different impacts on the spatial patterns and driving mechanisms of ANPP. The combination of climate change and human factors have resulted in an increase in the ANPP in the study area, having a greater influence than each of the two elements individually. In contrast, human activities have had a far greater impact on the decrease of ANPP than each of the two elements alone, being the main reason for the reduction in ANPP.

Based on the variation (increase/decrease) in ANPP within the research area, 20.44% of the region is influenced only by climate change, 35.99% of the region is influenced solely by human activities, and the remaining 42.79% of the area is influenced by both climate change and human activities (Figure 8). In Figure 8, ‘CDI’ stands for climate-dominated increase, ‘HDI’ for human-dominated increase, ‘CHI’ for climate- and human-dominated increase, ‘CDD’ for climate-dominated decrease, ‘HDD’ for human-dominated decrease, and ‘CHD’ for climate- and human-dominated decrease. Among them, from 2001 to 2018, in the area where ANPP increased, the increase caused by climate change accounted for 25.76%, which was mainly distributed around some mountains in the research area. Human activities accounted for an increase of 20.28%, concentrated in the southern part of Henan Province, the northern part of Hubei Province, central and southern Yunnan Province, central and northern Guizhou Province, central and northern Hunan Province, and central Jiangxi Province. The highest proportion, 53.97%, was influenced by both climate change and human activities, and was widely distributed across various regions of the study area. In the area where ANPP decreased, human activities accounted for an 83.44% decrease in this region, broadly distributed along the coastal provinces, whereas the combined influence of climate change and human activities only accounted for 5.43% and 11.14%, respectively.

3.5. Correlation between ANPP and Climate Factors and Heterogeneity Distribution on Elevation Gradient

According to Figure 9, from 2001 to 2018 in the study area, both the annual average temperature and total precipitation exhibited an upward trend. Figure 10 shows the distribution of the annual average temperature and total precipitation in the study area. The inter-annual variation in the annual average temperature was substantial, while the increase in total precipitation was relatively stable. The annual average temperature increased at a rate of 0.02 °C yr⁻¹, and the speed of precipitation increase was 14.90 mm yr⁻¹.

A partial correlation analysis was utilized to examine the relationship between climatic factors (annual total precipitation and annual average temperature) from 2001 to 2018 and ANPP in the study area. Figure 11 indicates the partial correlation coefficient between ANPP and temperature, where areas with a significant positive correlation were primarily located on both sides of the Daba Mountains spanning across Shaanxi and Sichuan provinces, both flanks of the Daliang Mountains and Wulian Peaks, and other mentioned regions. This suggests that in these areas, a rise in temperature facilitated the growth of ANPP, whereas it impeded it otherwise. Regions with a significantly negative correlation were more concentrated in specific zones mentioned, indicating that in these parts, a temperature increase was detrimental to ANPP growth, and a decrease was favorable instead.

Figure 11 represents the partial correlation coefficient between ANPP and precipitation. Interestingly, areas within the study region where the ANPP was significantly correlated with precipitation exhibited pronounced spatial differences. The areas with positive correlations primarily occupied the central, western, and northern parts of the study region, while negative correlations were concentrated in the southeastern coastal sections. Significant positive correlation areas predominantly occurred in specific mentioned mountainous and provincial regions, indicating that an increase in precipitation was beneficial to the enhancement of the ANPP in these locations, whereas it led to a decrease in the ANPP otherwise. Conversely, areas with a significantly negative correlation were mainly located in particular ecological protection zones and other mentioned areas, implying that increased precipitation in these regions was not conducive to ANPP growth, but a reduction in precipitation facilitates it.

The annual average temperature and total precipitation also exhibited different heterogeneous distributions at various altitudinal gradients. With the increase in altitude, both the annual average temperature and the total precipitation were in an overall declining trend (Figure 12). Regarding the correlation between ANPP and annual average temperature or total precipitation, if more than 50% of the pixels within a specific altitude range exhibited a positive (negative) correlation between ANPP and temperature (precipitation), it was determined that ANPP has a positive (negative) correlation with temperature (precipitation) in that range. Within 50 elevation bins ranging from 0 to 5000 m (at 100 m intervals), 82% of the bins showed a positive correlation between the ANPP and the annual average temperature, while 18% displayed a negative correlation, only occurring within the 1900 m to 2800 m elevation range. In comparison, bins showing a positive correlation between the ANPP and total precipitation constituted 56% of the bins, occurring within the 0 m to 2800 m elevation range, whereas 44% demonstrated a negative correlation, found within the 2800 m to 5000 m elevation range.

4. Discussion

4.1. Spatial and Temporal Distribution and Changing Trend of ANPP

Our research results indicate that from 2001 to 2018, the average ANPP across the entire subtropical research area in China exhibited a gradual increasing trend from north to south, a finding that aligns with previously reported trends [58,59,60]. The regional average ANPP stood at 677.17 gC m⁻² a⁻¹, coinciding with earlier research [28]. PNPP, standing for net primary productivity under conditions free from human disturbance, delineates the level of productivity of the most stable and mature vegetation that can be developed under natural conditions. This indicator is an essential tool for assessing the quality status of natural ecosystems, and it effectively distinguishes the impact of human activities on the ecological environment. In this study, the calculation of PNPP was based on the Thornthwaite memorial model. The HNPP, or human-affected net primary productivity, is calculated as the difference between the PNPP and the ANPP, quantifying the influence of human activities on vegetation productivity [52,53] The computational formulas for PNPP and HNPP are comprehensively detailed in Section 3.2. The findings revealed that the computed PNPP significantly exceeded both the ANPP and the HNPP, with the HNPP also being greater than the ANPP. The descending order of these indices was PNPP, HNPP, and ANPP. This concurred with previous research [49,61,62] and indicated the efficacy of these indicators in analyzing the study area.

The study area transitioned from the Central Plains to the southern part of China. The southern region is characterized by its richness in forest resources and a higher vegetation coverage compared to the central region. The abundance of precipitation, a suitable warm climate, and rich groundwater resources collectively contribute to favorable conditions for vegetation growth [63], resulting in a distribution trend where the vegetation is less dense in the north and denser in the south.

Areas with significant increases in the ANPP were detailed in the first section of Section 3 of this document. On a pixel scale, these regions of notable ANPP augmentation were predominantly found in the southeastern part of Yunnan Province, central Henan, the central and northern parts of Anhui, as well as near the mountains and the ecological function conservation areas within these provinces. The increase in the ANPP found within Yunnan Province was consistent with findings from previous studies [64,65]. There are mainly two reasons for this growth in ANPP in the area. The first reason is attributed to major national ecological restoration projects such as the ‘Grain for Green’ program, which led to the conversion of non-arboreal vegetation into forests in certain areas [66]. In addition, the retained farmlands, benefiting from favorable hydrothermal conditions, have also contributed to the increase in ANPP [65]. Anhui and Henan provinces are among the regions in China with a substantial distribution of farmlands, hosting a large number of croplands, where research by Luo and Xin et al. observed an increase in the area of croplands for wheat, rice, maize, etc. [67,68]; thus, the rapid increase in crop yield possibly spurred the increase in the region’s ANPP. The rise in mountainous ANPP can be attributed to the combined effects of temperature and precipitation, where, between altitudes of 1600 and 2800 m, photosynthesis in plants was positively correlated with temperature, a relationship that also applies to photorespiration and dark respiration [69]. Between 2001 and 2018, we observed an increasing trend in the total annual precipitation and the average annual temperature (Section 3.5). Abundant rainfall enriches the soil, providing more nutrients for the vegetation [69]. A naturally warm and humid environment offers improved hydrothermal conditions for mountain vegetation, promoting its growth. This accounts for the enhancement in the ANPP of the mountain vegetation. National policies can explain the significant increase in ANPP in the ecological function protection areas. Following the issuance of the “Outline of National Key Ecological Function Protection Area Planning” by the Chinese Ministry of Environmental Protection, ecological protection areas have become an indispensable part of environmental protection efforts in the country, especially the key ecological function protection areas, which are of great significance and highly valued. The designated protection areas in this study with significant ANPP increases were Southwest Karst Ecological Function Protection Area, Three Gorges Reservoir Area Ecological Function Protection Area, Qinling Mountain Ecological Function Protection Area, Dabie Mountain Ecological Function Protection Area, and Huai River Source Ecological Function Protection Area. These areas are categorized into two types: soil and water conservation ecological function protection areas and water source conservation ecological function protection. According to the “Outline of National Key Ecological Function Protection Area Planning”, both the soil and water conservation ecological function protection areas and water source conservation ecological function protection areas have established clear standards and guiding principles for vegetation protection. In the soil and water conservation ecological function protection areas, activities that destroy the forest, such as deforestation and slash-and-burn agriculture, as well as agricultural development on steep slopes, are prohibited. These measures aim to slow down vegetation degradation and maintain ecological stability. The area also proposed the need to implement soil and water conservation ecological restoration projects. For the water source conservation ecological function protection areas, the planning outline suggests integration with existing ecological conservation and development projects to enhance the comprehensive management and ecological restoration of forests, grasslands, and wetlands. Overall, these regulations strive to strengthen vegetation protection within these ecological function areas by limiting unsustainable land use activities and promoting ecological restoration projects. This way, not only is the ability to retain soil and water resources enhanced, but the net primary productivity of the vegetation within the region is also increased, thus promoting the long-term sustainability of the ecosystem.

In the middle and lower reaches of the Yangtze River and the Pearl River Delta urban agglomerations, we observed a significant decline in the ANPP, which is likely related to the high levels of urbanization and economic development in these areas [66,70]. Urbanization is characterized by the concentration of population and finance, along with an increase in artificial land cover, leading to a substantial reduction in vegetation [71]. In fact, as the middle and lower reaches of the Yangtze River and the Pearl River Delta are core areas of domestic economic development, their rapid economic growth and urban expansion have exerted significant pressure on the local vegetation. Therefore, the accelerating urbanization and economic activities have led to a notable decline in the ANPP in these regions.

It is noteworthy that there were differences in the annual average NPP (net primary productivity) among various forest types. In forests, evergreen broadleaf forests exhibited the highest annual average ANPP (728 gC m⁻² a⁻¹), followed by mixed forests (646 gC m⁻² a⁻¹). The other forest types were deciduous broadleaf forests (632 gC m⁻² a⁻¹) and evergreen needleleaf forests (622 gC m⁻² a⁻¹), respectively. These findings were consistent with previous research findings [72,73].

From a spatial perspective, evergreen needleleaf forests, deciduous needleleaf forests, and deciduous broadleaf forests are predominantly located in the western and northern parts of the study area. This distribution could be attributed to two main factors. Firstly, it may be due to a series of ecological forestry projects implemented in China since 1978, such as the Natural Forest Conservation Program and the Grain for Green Program [74,75]. These initiatives have promoted an increase in vegetation cover, forest area, and carbon storage. Secondly, from 2001 to 2018, there was an upward trend in the annual total precipitation and the average temperature in the study area. The increased rainfall enriched the soil fertility, providing more nutrients for vegetation, thus enhancing growth and leading to an increase in the ANPP [69]. Evergreen broadleaf forests are mainly distributed in the southeastern coastal regions of the study area and in southern Yunnan Province. In these areas, rapid urbanization has been a primary factor contributing to the decline in ANPP [76].

Different from the ANPP, which is influenced by human activities [77], the PNPP is primarily constrained by the topography, climate, and vegetation type [50]. In this study, the PNPP was significantly influenced by the temperature and precipitation, a phenomenon that can be fully explained from the perspective of algorithm construction [47]. The temporal variations of PNPP and ANPP were similar, yet the annual variation rate of PNPP was lower than that of ANPP, which might be attributed to the substantial deterioration of the vegetation level within the study area due to human activities, leading to a greater variation of ANPP relative to PNPP. This indicates that human activities have a significant impact on the variations of ANPP in this study area.

4.2. Variation of ANPP on Elevation Gradient

In the comprehensive analysis of altitudinal gradients, we observed a notable upward trend of ANPP within the study area as the elevation increased, gradually stabilizing with minor fluctuations before a significant decline; this finding was aligned with previous academic studies [65,78]. Additionally, the study area experienced a general decline in the average temperature and precipitation as the altitude rose (Figure 12) [65,79].

Specifically, in the elevation range of 0 to 400 m, vegetation gradually escaped the influence of human activities in the plains, showing an increasing trend in both abundance and area. This likely serves as a key factor for the significant rise in ANPP within this elevation interval. On the other hand, within the 400 to 1600 m elevation range, the impact of altitude on the average temperature and total precipitation was not significant due to their relatively minor fluctuations. Consequently, the ANPP exhibited variations in this range, but the overall trend remained relatively stable. In the altitude range of 1600 to 2800 m, we observed a trend in ANPP that first decreased and then increased with elevation. Within the elevation range of 1600 to 1900 m, the ANPP exhibited a declining trend, whereas between 1900 to 2800 m it showed an increasing trend. Notably, in the 1600 to 1900 m range, the ANPP primarily exhibited a positive correlation with the average temperature. However, this shifted to a predominantly negative correlation within the 1900 to 2000 m range and continued to demonstrate a negative correlation dominance throughout the 1900 to 2800 m range.

Within the altitude range of 1600 to 1900 m, the partial correlation between the temperature and the ANPP, although on the verge of shifting to negative, still presented as positive. In this elevation range, the temperature decreased with increasing altitude (Figure 12), and, concurrently, the ANPP showed a declining trend. This phenomenon could be attributed to the temperature decrease with rising altitude, which is unfavorable for the temperature conditions necessary for photosynthesis, thus limiting the rate of photosynthesis and carbon assimilation processes [78,80]. Therefore, the ANPP exhibited a downward trend under the influence of decreasing temperatures.

Further observation revealed that within the altitude range of 2000 to 2800 m, the partial correlation between temperature and ANPP was negatively correlated. In this elevation range, the temperature continued to decrease with increasing altitude, while the ANPP values showed an upward trend. This trend might be related to the increase in solar radiation within this altitude range. According to previous research, at these altitudes, the positive influence of solar radiation on the ANPP progressively intensifies, exhibiting a significant positive driving effect [65]. Consequently, within this elevation range, the favorable impact of solar radiation on ANPP may surpass the negative effects of decreasing temperature, ultimately leading to an increase in the ANPP, which peaked within this altitude range.

In regions above 2800 m, the annual average temperature dropped below 10 °C and precipitation also fell below 1100 mm, but still remained above 650 mm a⁻¹ 100m⁻¹. Both the average temperature and precipitation exhibited a decreasing trend with increasing altitude. The ANPP showed a significant downward trend, consistent with previous research findings [65]. In areas above 5000 m, the ANPP remained below 5 gC m⁻² a⁻¹.

In line with previous studies [51,81,82,83], we observed a negative response of ANPP to precipitation in high altitude areas, potentially due to the fact that in subtropical regions, despite precipitation decreasing with lower altitudes, the values remain between 600 and 800 mm a⁻¹ 100m⁻¹ (with a mean value of 1271.27 mm a⁻¹ 100 m⁻¹ in low altitude areas), possibly exceeding vegetation growth requirements. When precipitation surpasses the vegetation growth demands, the photosynthesis of vegetation reacts adversely to the reduced radiation and increased relative humidity. Hence, from low to high altitude areas, as precipitation values approach the vegetation growth needs, the adverse reactions induced by radiation and relative humidity decrease, causing an increment in the ANPP values. However, the influencing factors for ANPP in high altitude areas are not limited to precipitation; studies demonstrate that temperature remains the dominant factor for ANPP in these regions [84,85]. In this study, the ANPP in high altitude areas predominantly showed a positive correlation with temperature. Consequently, despite the reduction in precipitation with increased altitude causing some increase in the ANPP, the overall trend of ANPP was significantly downward due to the dominant effect of temperature. Within elevation intervals divided into 100 m units, different factors dictated the changes in ANPP at various altitudes. The variation in ANPP along the elevation gradient illustrated the influence of several meteorological elements, such as precipitation, temperature, and solar radiation, coupled with human activities, on the altitudinal distribution of ANPP.

4.3. The Relative Effects of Climate Change and Human Activities on ANPP

The alteration in ANPP is primarily induced by human activities and climate fluctuations. Numerous preceding studies have utilized ANPP, HNPP, and PNPP to differentiate these two elements’ effects on ANPP [18,48,49,61,62,86]. Our research findings indicate that a combined influence of climate change and human activities has led to the increase in ANPP in the study area, with their combined effect being greater than the individual impacts of climate change or human activities on the vegetation’s ANPP. This increase can be ascribed to the escalating temperatures and heightened precipitation in the region, which create conducive climatic conditions in most areas. These factors boost photosynthesis and carbon storage in vegetation, ultimately contributing to the elevation of ANPP [87,88]. Additionally, human-driven efforts towards vegetation restoration, including grassland conservation policies and diverse ecological restoration initiatives, have markedly enhanced vegetation coverage and alleviated the detrimental effects of human activities on ANPP [89,90]. Thus, in most of the areas within the study region where there is an increase, the growth in ANPP is dominantly driven by both climate change and human activities [62].

The increase in ANPP in certain mountainous regions of the study area can be attributed solely to climate change. This positive effect is likely due to two factors: the lack of human activity in these secluded mountainous regions and the favorable conditions created by increased temperatures for vegetation growth. Temperature is a critical factor that limits plant growth. Research has indicated [88,91] that in areas with an elevation of approximately 1000 m, photosynthesis in vegetation is often hindered at temperatures below 20 °C, which results in a lower ANPP in many mountainous areas due to the cooler climate. Nevertheless, as per the research presented in Section 3.5 of this study, the study region experienced a general upward trend in annual average temperatures from 2001 to 2018. The gradual increase in temperature promotes vegetation growth [89,92], especially in mountainous areas, leading to a positive influence of climate change in these regions.

Regions where solely human interventions have resulted in a rise in the ANPP were predominantly concentrated in southern Henan Province, northern Hubei Province, northern Anhui Province, central-southern Yunnan Province, and central-northern Guizhou Province. These areas are primarily characterized by farmlands and the karst regions of China. In these regions, the impact of climate change on ANPP is relatively minor compared to human activities. As analyzed in Section 4.1, the areas of southern Henan, northern Hubei, and northern Anhui are abundant in farmlands, where human activities, closely linked to agricultural practices, play a significant role. Factors such as expanded farmland area or enhanced efficiency of land use are largely due to human involvement [67,68]. In Yunnan and Guizhou Provinces, located in the Southwest China karst region, ecological protection projects, ecological migration, and initiatives such as converting farmland back to forest or grassland have created favorable conditions for the accumulation of ANPP [93]. Numerous studies have revealed an increase in vegetation cover and a growing trend of ANPP in the Southwest China karst region [89,93,94,95]. As a result, human interventions have resulted in a rise in ANPP in these areas, exerting a more significant effect than that of climate change.

In coastal areas where ANPP has declined, human activities have been the primary factor driving the reduction, with their negative impact being more significant than that of climate change, aligning with the findings of previous research [76,87,88]. This is likely due to the rapid urbanization in these regions. Studies have shown that accelerated urbanization leads to a deterioration in the quality of forest ecosystems and a significant decrease in vegetation cover. Urban expansion encroaches upon the habitats vital for vegetation, resulting in substantial reductions in vegetation cover and ANPP. Although previous research indicates [87] that in highly urbanized areas, low productivity vegetation is gradually replaced by higher productivity types, the increase in ANPP due to this vegetation transformation was insufficient to offset the decline caused by the expansion of impervious urban surfaces. Moreover, according to a study by Xin et al., the adverse effects of nocturnal warming on vegetation growth in the Fujian and Guangdong regions also contributed to the decline in ANPP [76]. Therefore, in these regions, human activities predominantly contribute to the reduction in ANPP.

4.4. Correlation between Climate Factors and ANPP

Temperature and precipitation have always been the two most commonly used factors in studying the driving models between ANPP and meteorological elements [49,64,76,96,97,98,99], as they largely affect the growth and distribution of vegetation. However, previous studies have not reached consistent conclusions regarding the impact of these two meteorological factors on ANPP. Studies reporting both positive and adverse effects of these factors on ANPP have been documented [48,62,65,96]. This study largely agrees with the conclusion that temperature has a positive effect on ANPP, with 65.14% of the regions showing a positive correlation between temperature and ANPP, and a trend of temperature increase was observed in areas with positive correlations. Therefore, in these regions, the increase in temperature leading to a rise in ANPP might be due to the fact that, under the global warming trends in the study area, a temperature increase before reaching the optimum growth temperature can enhance soil microbial activity, improve photosynthetic efficiency, and promote organic matter accumulation in plants [97,98], thereby improving productivity. Another possible reason is that the rise in temperature advances the phenological phase of the vegetation, extending the growth period and accelerating the photosynthesis process, which promotes the carbon assimilation process and biomass accumulation [100]. However, in areas within the research region where the temperature was negatively correlated with ANPP, these were mainly distributed in the central and northern parts of Yunnan Province, the central high altitude areas of Taiwan Province, and the central and southern parts of Hubei Province.

We observed a trend of temperature increase in the middle and upper areas of Yunnan Province and the central high-elevation areas of Taiwan Province, implying that in these areas, the rise in temperature has a negative effect on ANPP, leading to a decrease in ANPP. This might be because global warming increases the evapotranspiration of the ecosystem, which intensifies soil water shortage, restricting vegetation growth and reducing ANPP [76,99]. Another part of the reason could be that in high altitude areas, global warming causes snow and ice to melt, increasing soil moisture at plant root zones, causing plants to be in an anaerobic state, which in turn leads to a decrease in ANPP [101]. In the southern part of Hubei Province, we observed a trend of temperature decrease, which in this area promotes vegetation growth, resulting in an increase in ANPP. The Han River Plain, located in the heart of Hubei Province and extending southwards, encompasses a vast expanse of farmland and stands as one of the key grain-producing bases. Studies have shown that the arable land area in the Han River Plain is gradually expanding [67,68], and with the continuous upgrading of agricultural technology, its yield is expected to increase. The rapid increase in crop yield is likely the primary reason for the increase in ANPP in this area, and its impact outweighs the negative effects of a temperature decrease on ANPP, possibly being the reason why the ANPP increased in this region despite the downward trend in temperature.

Changes in precipitation can influence the growth of plant root systems and the supply of moisture [102]. In our study, precipitation was found to be a more significant enhancer of ANPP, as approximately 60% of the regions exhibited a positive correlation between precipitation and ANPP. These areas have experienced a yearly rise in precipitation, suggesting that the increase in rainfall positively influences vegetation growth in these regions. Precipitation represents the maximum amount of water available to regional vegetation [51]. The level of precipitation directly affects the soil moisture content, which is a key factor linking precipitation and ANPP. Precipitation drives dynamic changes in soil inorganic nitrogen within the system because the main source of new nitrogen in the system is net deposition. The increase in precipitation enhances soil moisture content, promoting plant activities and improving the efficiency of photosynthesis in vegetation, allowing for the accumulation of more organic matter [18]. However, the remaining 40% of areas with a negative correlation and those with a positive correlation exhibited considerable spatial heterogeneity; the former were often found in the eastern coastal areas closer to the sea, while the latter were more often distributed inland. In coastal areas, we also observed a trend of increasing precipitation; the negative correlation led to a decrease in ANPP in these regions. This might be because the increase in precipitation was accompanied by an increase in cloud cover, thereby reducing solar radiation, altering the oxygen environment in the root system, suppressing the photosynthesis of vegetation, and leading to a decline in productivity [82,83]. An increase in precipitation can also lead to a reduction in soil organic matter, accelerating the rate of soil erosion and causing flood disasters, destroying the environment for vegetation growth, and reducing ANPP [103].

4.5. Limitations

In this study, we investigated the distribution of net primary productivity (ANPP) along an altitudinal gradient and its driving effects in subtropical China. Subtropical forest ecosystems are complex and diverse, and are influenced by a variety of factors. Temperature and precipitation are the key factors affecting ANPP, and other factors, such as CO₂ emissions, nitrogen deposition, and human activities, also play important roles, but these factors were not addressed in this paper.

5. Conclusions

In this paper, the EC-LUE model and CUE were used to simulate and analyze the ANPP in subtropical regions of China from 2001 to 2018 and its heterogeneity distribution at different altitudes. The Thornthwaite memorial model was used to calculate the potential net primary productivity of vegetation, and based on this scenario an analysis was set up to quantitatively assess the impacts of climate change and human activities on vegetation productivity in the study area. Finally, the main driving effects of climate factors on ANPP were studied. The conclusions are as follows:

(1)

From 2001 to 2018, the ANPP in this region showed an overall growth trend, especially in mountainous areas, protected areas, and cultivated areas in the northeast plain, but showed an obvious downward trend around the southeast coastal urban agglomerations.

(2)

Human activities and climate change are the main factors driving the ANPP changes in the study area, and they jointly dominated the ANPP changes in 42.79% of the areas, in which the single impact of human activities exceeded the single impact of climate change on ANPP, dominating 35.99% and 20.44% of the areas, respectively.

(3)

ANPP presented a pattern of significant rise, slight fluctuation, and significant decline along the altitudinal gradient, and the climate-driven effects on ANPP were different at different altitude gradients.

(4)

Temperature and precipitation significantly affected the growth and distribution of vegetation in this region, showing an overall positive correlation with ANPP, but this relationship showed significant heterogeneity among different regions.

In future studies, we will use more climatic and human factors related to ANPP to participate in the analysis of ANPP, and combine these factors to make a more accurate and comprehensive analysis of the driving effect of ANPP and the heterogeneity of changes along the elevation gradient in subtropical regions, so as to have a more comprehensive understanding of the carbon balance of subtropical forest ecosystems.