Net Primary Productivity Variations Associated with Climate Change and Human Activities in Nanjing Metropolitan Area of China

Rapid economic development has changed land use and population density, which in turn affects the stability and carbon sequestration capacity of regional ecosystems. Net primary productivity (NPP) can reflect the carbon sequestration capacity of ecosystems and is affected by both climate change and human activities. Therefore, quantifying the relative contributions of climate change and human activities on NPP can help us understand the impact of climate change and human activities on the carbon sequestration capacity of ecosystems. At present, researchers have paid more attention to the impact of climate change and land use change on NPP. However, few studies have analyzed the response of the NPP to gross domestic product (GDP) and population density variations on a pixel scale. Therefore, this paper analyzes the impact of climate change and human activities to NPP on a pixel scale in the Nanjing metropolitan area. During the period 2000–2019, the annual mean NPP was 494.89 g C·m−2·year−1, and the NPP in the south of the Nanjing metropolitan area was higher than that in the north. The NPP was higher in the forest, followed by unused land, grassland, and cropland. In the past 20 years, the annual mean NPP showed a significant upward trend, with a growth rate of 3.78 g C·m−2·year−1. The increase in temperature and precipitation has led to an increasing trend of regional NPP, and the impact of precipitation on NPP was more significant than that of temperature. The transformation of land use from low-NPP type to high-NPP type also led to an increase in NPP. Land use change from high-NPP type to low-NPP type was the main cause of regional NPP decline. Residual analysis was used to analyze the impact of human activities on NPP. Over the last 20 years, the NPP affected by human activities (NPPhum) showed a high spatial pattern in the south and a low spatial pattern in the north, and the annual mean NPPhum also showed a fluctuating upward trend, with a growth rate of 2.00 g C·m−2·year−1. The NPPhum was influenced by both GDP and population density, and the impact of population density on NPP was greater than that of GDP.


Introduction
The sixth assessment report of the United Nations' Intergovernmental Panel on Climate Change (IPCC) shows that over the past 20 years, global surface temperatures had been warmer in each decade than in the previous decade [1]. Global warming leads to rising sea levels and extreme weather events. Studies have shown that the rising concentration of atmospheric carbon dioxide (CO 2 ) is the main driving factor of global warming. Vegetation can fix atmospheric CO 2 through photosynthesis, thereby mitigating global warming. However, in economically developed regions, rapid economic development has led to an increase in urban land construction and the agglomeration of population, which in turn reduces the area of forest land, grassland, and cropland. Therefore, climate change and human activities have affected the stability and biodiversity of terrestrial ecosystems, thereby degrading the carbon sequestration capacity. Quantifying the impact of climate NPP is the NPP affected only by climate change, while the observed NPP is affected by climate change and human activities. Therefore, the residual is the NPP affected by human activities. In this paper, the observed NPP was obtained from the MODIS MOD17A3HGF NPP dataset, and the potential NPP was estimated by the Thornthwaite Memorial model. The Thornthwaite Memorial model, which estimates the potential NPP from a statistical regression relationship between potential NPP and the climatic factors of air temperature and precipitation, is widely used in the estimation of potential NPP [16,31,33].
The Nanjing metropolitan area is a region with a high level of economic development and high population density in China. The rapid economic development has brought great pressure on the security of ecosystems in the region. In the past 20 years, the area of forest, grassland, and cropland had decreased by 3192, 75, and 58 km 2 , respectively. Therefore, analyzing the impact of climate change and human activities on the vegetation NPP in the Nanjing metropolitan area is helpful for understanding the impact of climate change and human activities on the carbon sequestration capacity of ecosystems, and provides theoretical reference and data support for the government to understand the regional carbon budget and formulate carbon emission reduction strategies.
Based on the above scientific issues, this paper seeks to answer the following questions: (1) What are the temporal and spatial variation patterns of NPP in the Nanjing metropolitan area? (2) How do climate change and human activities affect NPP?

Study Area
The Nanjing metropolitan area, as the first inter-provincial metropolitan area constructed in China and the first metropolitan area approved by the state, is located between 117 • 9 -119 • 57 E and 29 • 57 -34 • 5 N. It is the core area of the Yangtze River delta urban agglomeration of China and spans the two provinces of Jiangsu and Anhui. The cities in it include Nanjing, Zhenjiang, Yangzhou, Huaian, Maanshan, Chuzhou, Wuhu, Xuancheng, Jintan, and Liyang, with a total area of 66,000 km 2 (see Figure 1a). The Nanjing metropolitan area has a subtropical monsoon climate, with abundant moisture and heat. The average annual temperature varies between 15 and 22 • C, and the annual precipitation ranges from 800 to 1600 mm. With the continuous increase in the urbanization level of the Nanjing metropolitan area, a large amount of cropland, forest, and grassland had been occupied by urban land (see Figure 1b,c). During the period 2000-2019, the area of forest, grassland, and cropland had decreased by 3192, 75, and 58 km 2 , respectively. The urban land had increased significantly, mainly from the conversion of forests.

Data Sources and Preprocessing
Four types of datasets were used in this paper, and these are (1) the NPP dataset; (2) meteorological datasets, including temperature and precipitation; (3) the datasets of socio-economic and demographic, such as GDP and population density; and (4) land cover maps. The MODIS MOD17A3HGF NPP dataset was obtained from NASA's Land Processes Distributed Active Archive Center (LP DAAC, https://lpdaac.usgs.gov/data_ access/data_pool, accessed on 15 June 2022). The NPP dataset was resampled to match a spatial resolution of 1km using ArcGIS (version 10.5). The meteorological datasets were downloaded from the National Earth System Science Data Center of China (http: //www.geodata.cn/index.html, accessed on 15 June 2022). The temperature and precipitation datasets were verified by the data from 496 meteorological observation stations in China [34]. The land cover maps, GDP, and population density datasets were downloaded from the Resource and Environmental Science and Data Centre of the Chinese Academy of Sciences (http://www.resdc.cn/Default.aspx, accessed on 15 June 2022). The spatial resolution of these datasets is 1 km. The GDP and population density datasets were seen in the Supplementary Data.

Data Sources and Preprocessing
Four types of datasets were used in this paper, and these are (1) the NPP da meteorological datasets, including temperature and precipitation; (3) the datasets economic and demographic, such as GDP and population density; and (4)

Estimation of Potential NPP
The potential NPP of vegetation (NPP pot ) is the NPP of the undisturbed and humanimpacted ecosystem and is only influenced by climatic factors. The Thornthwaite Memorial model, which estimates the NPP pot from a statistical regression relationship between NPP pot and climatic factors of air temperature and precipitation [35], has been widely used to calculate the NPP pot [16,31,33]. The calculation equations are as follows: (1) v = 1.05r (2) where v is the actual annual evapotranspiration (mm), L is the average annual evapotranspiration (mm), t is the average annual temperature ( • C), and r is the total annual precipitation (mm).

Estimation of the NPP Affected by Human Activities
The observed NPP (NPP act ) is affected by climate change and human activities. Therefore, the NPP affected by human activities (NPP hum ) can be calculated by the difference between NPP pot and NPP act , and the equation is: where the NPP act was obtained from the MODIS MOD17A3HGF NPP dataset. If NPP hum > 0, it means that human activities have a negative effect on NPP, which caused the NPP act to be smaller than NPP pot . If NPP hum < 0, it means that human activities have a positive effect on NPP.

Slope Trend Analysis
Linear regression analysis based on the least-squares method was used to calculate the interannual change trends of NPP in the Nanjing metropolitan area. The slope of the linear regression equation can be calculated as: where n is the number of years, and NPP i is the NPP in the year i. Slope is the interannual variation rate of NPP. If Slope > 0, it means that the NPP shows an increasing trend. Conversely, it indicates that NPP shows a decreasing trend [36].

Correlation Analysis
The Pearson correlation coefficient was used to analyze the relationship between NPP and meteorological factors and the relationship between NPP and human activities factors. The Pearson correlation coefficient can be calculated as: where R xy is the correlation coefficient between the two variables, x i represents the annual average temperature or precipitation, GDP, or population density in the year i, y i represents the annual average NPP in the year i, and x and y represent the mean value of x and y, respectively. If R xy > 0, it means that x is positively correlated with y, and if R xy < 0, it indicates that x is negatively correlated with y. The Student's t-test was also used to test the significance level of the correlation between NPP and GDP, and the correlation between NPP and population density. The Pearson correlation coefficients and their significance levels were classified into nine categories (Table 1).

Spatial Pattern of the NPP act
During the period 2000-2019, the mean total value of NPP act was 30.69 Tg C·year −1 in the Nanjing metropolitan area, while the mean NPP act per unit was 494.89 g C·m −2 ·year −1 . Figure 2 shows that the NPP act in the southern regions of the Nanjing metropolitan area were higher than those in the other regions. Figure 1 shows that the forest is widely distributed in the southern regions, accounting for 59.46% of the area of Xuancheng. Therefore, the NPP act were mostly above 600 g C·m −2 ·year −1 in the southern regions. The area of regions with low NPP act (NPP act < 300 g C·m −2 ·year −1 ) accounted for 1.38% of the area of the Nanjing metropolitan area, and these regions were mainly distributed around rivers and urban land.

NPPact Interannual Variability
Over the last 20 years, the annual NPPact in the Nanjing metropolitan area had generally shown a fluctuating upward trend, increasing from 440.97 in 2000 to 497.71 g We extracted the regions where land use types have not changed in the past 20 years and estimated the mean annual NPP for different vegetation types. The NPP in the forest was higher, with a mean value of 585.61 g C·m −2 ·year −1 , followed by the unused land (535.17) and grassland (500.80). The lowest NPP appeared in the cropland, with a mean value of 480.29 g C·m −2 ·year −1 .

NPP act Interannual Variability
Over the last 20 years, the annual NPP act in the Nanjing metropolitan area had generally shown a fluctuating upward trend, increasing from 440.97 in 2000 to 497.71 g C·m −2 ·year −1 in 2019 (see Figure 3a). The annual growth rate was 3.78 g C·m −2 ·year −1 . The annual NPP act varied between 440.97 and 569.94 g C·m −2 ·year −1 , with a coefficient of variation of 6.8%. The lowest value of NPP act appeared in 2000, while the highest value occurred in 2014. Figure 3 also shows that the annual NPP act was affected by the temperature and precipitation. The increase in temperature and precipitation in the past 20 years had led to an overall upward trend of regional vegetation NPP act . The correlation between the NPP act and precipitation was large, with a correlation coefficient of 0.45. The correlation between the NPP act and temperature was weak, and the correlation coefficient was 0.17. It means that in the Nanjing metropolitan area, the impact of the precipitation on NPP act was greater than that of temperature. The continuous hydrological drought appeared from 2004 to 2008 in the lower reaches of the Yangtze River [37], precipitation decreased, and this led to the lower annual NPP act for the period. The continuous decrease in temperature from 2008 to 2011 led to a significant downward trend of NPP act . The NPP act decreased from 2015 to 2017, and this may be caused by the downward trend of precipitation.   Figure 4 shows that in the past 20 years, the NPPact in most regions of the Nanjing metropolitan area had shown an increasing trend, accounting for 81.80% of the area of the Nanjing metropolitan area. In these regions, 43.24% of the regional land use types changed from low-NPP type to high-NPP type. It means that the transformation of land use from  Figure 4 shows that in the past 20 years, the NPP act in most regions of the Nanjing metropolitan area had shown an increasing trend, accounting for 81.80% of the area of the Nanjing metropolitan area. In these regions, 43.24% of the regional land use types changed from low-NPP type to high-NPP type. It means that the transformation of land use from low-NPP type to high-NPP type led to an increase in NPP. There was about 48.87% of the area of the Nanjing metropolitan area where the growth rate of the NPP act was greater than 3.78 g C·m −2 ·year −1 . There was about 6.88% of the area of the Nanjing metropolitan area where the NPP act had shown a downward trend. In these regions, 69.42% of the regional land use types changed from high-NPP type to low-NPP type. It means that the land use change from high-NPP type to low-NPP type was the main cause of regional NPP decline. There was about 2.85% of the area of the Nanjing metropolitan area where the decline rate of the NPP act was greater than 3.78 g C·m −2 ·year −1 , and most of them were located in the southwest of Nanjing, southeast of Maanshan, and the riverside of Yangtze River in Nanjing and Zhenjiang. In those regions, the area of urban land had grown rapidly since the 2000s.

Spatial Pattern of NPPhum
Based on Equation (4), this study estimated the NPP affected by human activi (NPPhum) in the Nanjing metropolitan area from 2000 to 2019. Figure 5 shows that NPPhum in the southern regions were higher than those in the northern regions. It me that the impact of human activities on NPP in the southern regions was greater than t

Spatial Pattern of NPP hum
Based on Equation (4), this study estimated the NPP affected by human activities (NPP hum ) in the Nanjing metropolitan area from 2000 to 2019. Figure 5 shows that the NPP hum in the southern regions were higher than those in the northern regions. It means that the impact of human activities on NPP in the southern regions was greater than that in the northern regions. The average annual NPP hum was 917.23 g C·m −2 ·year −1 , and the minimum value of NPP hum was 528.64 g C·m −2 ·year −1 , indicating that human activities in the Nanjing metropolitan area mainly had a negative impact on NPP. The regions with high values of NPP hum (NPP hum > 1100 g C·m −2 ·year −1 ) accounted for 7.64% of the study area, mainly distributed in Wuhu, southwest of Xuancheng, and the riverside of the Yangtze River in Nanjing and Maanshan, while the regions with low values of NPP hum (NPP hum < 800 g C·m −2 ·year −1 ) accounted for 20.33% of the study area, which was mainly located in Huaian, northern Chuzhou, and northern Yangzhou.

NPPhum Interannual Variability
Over the last 20 years, the annual NPPhum in the Nanjing metropolitan erally shown a fluctuating upward trend, with an annual growth rate of 2.0 (see Figure 6). The annual NPPhum varied between 763.76 and 1103.07 g C·m a coefficient of variation of 10.4%. It indicates that human activities had a on NPP, which caused the NPPact to be smaller than NPPpot. The maximum in 2003, while the minimum value appeared in 2004. During the period

NPP hum Interannual Variability
Over the last 20 years, the annual NPP hum in the Nanjing metropolitan area has generally shown a fluctuating upward trend, with an annual growth rate of 2.0 g C·m −2 ·year −1 (see Figure 6). The annual NPP hum varied between 763.76 and 1103.07 g C·m −2 ·year −1 , with a coefficient of variation of 10.4%. It indicates that human activities had a negative effect on NPP, which caused the NPP act to be smaller than NPP pot . The maximum value occurred in 2003, while the minimum value appeared in 2004. During the period 2004-2013, the NPP hum was low. This means that the negative effects of human activities on NPP had been alleviated. However, after 2013, the NPP hum increased rapidly. This means that the negative effects of human activities on NPP began to increase and reached their peak in 2016. After 2016, the NPP hum decreased significantly.

NPPhum Interannual Variability
Over the last 20 years, the annual NPPhum in the Nanjing metropolitan area has erally shown a fluctuating upward trend, with an annual growth rate of 2.0 g C·m −2 ·y (see Figure 6). The annual NPPhum varied between 763.76 and 1103.07 g C·m −2 ·year −1 , a coefficient of variation of 10.4%. It indicates that human activities had a negative e on NPP, which caused the NPPact to be smaller than NPPpot. The maximum value occu in 2003, while the minimum value appeared in 2004. During the period 2004-2013 NPPhum was low. This means that the negative effects of human activities on NPP been alleviated. However, after 2013, the NPPhum increased rapidly. This means tha negative effects of human activities on NPP began to increase and reached their pea 2016. After 2016, the NPPhum decreased significantly.

Spatial Variations of NPP hum
During the period 2000-2019, the NPP hum in most regions of the Nanjing metropolitan area showed an increasing trend, covering 58.43% of the total area of the Nanjing metropolitan area (see Figure 7). In these regions, the negative effects of human activities on NPP had been increasing, and 54.46% of the regional land use types changed from high-NPP type to low-NPP type. This means that the transformation of land use from low-NPP type to high-NPP type led to an increase in NPP hum . There was about 41.82% of the regions where the growth rate of NPP hum was greater than 2 g C·m −2 ·year −1 , and these regions were mainly located in southeastern Chuzhou, central and southern Yangzhou, Nanjing, Zhenjiang, Maanshan, Jintan, Liyang, Wuhu, and Xuancheng. The area of the regions where the NPP hum decreased accounted for 30.98% of the total area of the Nanjing metropolitan area, and these regions were mainly distributed in Huaian, central and northern Yangzhou, and northwest of Chuzhou. In these regions, the negative effects of human activities on NPP had been alleviated, and 40.32% of the regional land use types changed from low-NPP type to high-NPP type. It means that the land use change from low-NPP type to high-NPP type led to a decrease in NPP hum . The area with the decline rate of NPP hum greater than NPP type led to a decrease in NPPhum. The area with the decline rate of NPPhum grea than 2 g C·m −2 ·year −1 accounted for 16.00% of the total area of the Nanjing metropoli area, mainly distributed in Huaian and northwest of Chuzhou. This may be caused by "Grain for Green Project" which has been implemented to protect natural forest resour since 1998, and the area of forest increased by about 483 and 133 km 2 in Chuzhou a Huaian during 2000-2019.

Impact of Climatic Factors on NPP act
Climate change plays an important role in changing the terrestrial ecosystem NPP [14], while temperature and precipitation are the most direct and sensitive factors affecting the NPP [15,20]. In this paper, the Granger causality analysis was performed in Stata (version 13) to demonstrate the direction of causality between NPP act and climatic factors (see Table 2). The Chi-square as well as the associate probability show a rejection of the null hypothesis that temperature does not cause the variation of NPP act , while the Chi-square does not allow us to reject the null hypothesis that NPP act does not cause the variation of temperature. Table 2 also shows that the existence of a causal relation between precipitation and NPP act is evident. This means that in the Nanjing metropolitan area, the temperature and precipitation both affected the NPP act . The Pearson correlation coefficients between NPP act and temperature ranged from 0.60 to 0.76 (see Table 3). The area of the regions where the correlation between NPP act and temperature was a moderate or high positive correlation (R ≥ 0.3) accounted for 16.02% of the total area of the Nanjing metropolitan area. These regions were mainly distributed in northern Nanjing, central and eastern Chuzhou, eastern Huaian, and southern Yangzhou (see Figure 8a). The area of the regions with a mainly weak correlation (−0.3 < R < 0.3) between NPP act and temperature accounted for 82.01% of the study area. Table 2 also shows that the correlation coefficients between NPP act and precipitation ranged from −0.68 to 0.82, with 60.44% of the study area having a moderate or high positive correlation (R ≥ 0.3). Among them, the area of the regions where the correlation between NPP act and precipitation was a moderate positive correlation (0.3 ≤ R < 0.8) accounted for 60.43% of the study area and was mainly located in Chuzhou, Maanshan, Xuancheng, Nanjing, Zhenjiang, Jintan, and Liyang (see Figure 8b). The results indicate that the NPP act was more sensitive to precipitation in the Nanjing metropolitan area. Compared to temperature, precipitation was the dominant factor affecting the NPP act in the Nanjing metropolitan area. Table 3. Correlation coefficients (R) and proportion of the area in the Nanjing metropolitan area between the NPP act and climatic factors.   Irrigation on cropland also affected the changes in NPP act . We extracted the slope of NPP act where the land use type of cropland had not been changed during the period of 2000-2019 (see Figure 9a), and the results show that in 93.66% of the cropland the NPP act showed an increasing trend, while in only 6.34% of the cropland the NPP act showed a decreasing trend. Figure 9b shows that the irrigation area of 10 cities has increased, with an increasing area of 6177.21 km 2 . Therefore, the vigorous promotion of irrigation in recent years is indeed conducive to the increase of NPP.

Impact of GDP on NPPhum
Previous studies have shown that GDP influenced regional vegetation NPP, with significant negative correlations distributed in economically developed areas [27]. Growth in GDP significantly deepens vegetation degradation [38], and the GDP was significantly

Impact of GDP on NPP hum
Previous studies have shown that GDP influenced regional vegetation NPP, with significant negative correlations distributed in economically developed areas [27]. Growth in GDP significantly deepens vegetation degradation [38], and the GDP was significantly negatively correlated with NPP in southeastern China [39]. Based on Equation (4) and Table 1, this paper estimated the correlation coefficient and its significance level between NPP hum and GDP (see Figure 10). The results show that in the past 20 years, the correlation coefficients between NPP hum and GDP in the Nanjing metropolitan area ranged from −1 to 1 and were mainly weakly correlated (−0.3 < R < 0.3), accounting for 49.17% of the study area (see Figure 10a). The area of the regions where the correlation between NPP hum and GDP was a moderate or high positive correlation (R ≥ 0.3) accounted for 38.96% of the study area. These regions were mainly distributed in Xuancheng, Wuhu, north and south of Nanjing, Liyang, eastern Zhenjiang, and southeastern Yangzhou. The regions with moderate and high negative correlation (R ≤ −0.3) between NPP hum and GDP accounted for 11.87% of the study area, mainly distributed in Huaian and Chuzhou. In these regions, the area of forest showed an increasing trend during 2000-2019.
GDP was a moderate or high positive correlation (R ≥ 0.3) accounted for 38.96% of the study area. These regions were mainly distributed in Xuancheng, Wuhu, north and south of Nanjing, Liyang, eastern Zhenjiang, and southeastern Yangzhou. The regions with moderate and high negative correlation (R ≤ −0.3) between NPPhum and GDP accounted for 11.87% of the study area, mainly distributed in Huaian and Chuzhou. In these regions, the area of forest showed an increasing trend during 2000-2019. Figure 10. The spatial patterns of (a) the correlation coefficient between NPPhum and GDP, (b) and significant test of these data.
The regions with a significant positive correlation (R ≥ 0.3, p < 0.05) between NPPhum and GDP accounted for 7.63% of the study area. These regions were mainly distributed in southwestern Xuancheng, central and eastern Maanshan, and eastern Zhenjiang (see Figure 10b). Among them, the regions with a highly significant positive correlation (R ≥ 0.3, p < 0.01) between NPPhum and GDP accounted for 5.22% of the study area, mainly distributed in southeastern Xuancheng and central Maanshan. Figure 10b also shows that the regions with a significant negative correlation (R ≤ −0.3, p < 0.05) between NPPhum and GDP accounted for 1.51% of the study area. These regions were mainly distributed in central Chuzhou, central Huaian, southern Yangzhou, and northern Zhenjiang. Among them, the regions with a highly significant negative correlation (R ≤ −0.3, p < 0.01) between NPPhum and GDP accounted for 0.42% of the study area, mainly distributed in central Chuzhou and southern Yangzhou. Figure 10. The spatial patterns of (a) the correlation coefficient between NPP hum and GDP, (b) and significant test of these data.
The regions with a significant positive correlation (R ≥ 0.3, p < 0.05) between NPP hum and GDP accounted for 7.63% of the study area. These regions were mainly distributed in southwestern Xuancheng, central and eastern Maanshan, and eastern Zhenjiang (see Figure 10b). Among them, the regions with a highly significant positive correlation (R ≥ 0.3, p < 0.01) between NPP hum and GDP accounted for 5.22% of the study area, mainly distributed in southeastern Xuancheng and central Maanshan. Figure 10b also shows that the regions with a significant negative correlation (R ≤ −0.3, p < 0.05) between NPP hum and GDP accounted for 1.51% of the study area. These regions were mainly distributed in central Chuzhou, central Huaian, southern Yangzhou, and northern Zhenjiang. Among them, the regions with a highly significant negative correlation (R ≤ −0.3, p < 0.01) between NPP hum and GDP accounted for 0.42% of the study area, mainly distributed in central Chuzhou and southern Yangzhou.

Impact of Population Density on NPP hum
The increase in population density can lead to the impact of human activities on terrestrial ecosystems becoming more and more severe [40], thereby degrading the regional vegetation NPP. Some studies have found that population density has a strong negative correlation with NPP, especially in densely populated areas [27]. An increase in population density significantly deepens vegetation degradation [38]. In the past 20 years, the correlation coefficients between NPP hum and population density in the Nanjing metropolitan area ranged from −1 to 1, mainly showing a moderate negative correlation (−0.8 < R ≤ −0.3), accounting for 36.75% of the study area (see Figure 11a). The area of the regions where the correlation between NPP hum and population density was a moderate or high positive correlation (R ≥ 0.3) accounted for 25.87% of the study area. These regions were mainly distributed in northwestern Huaian, the north-central part of Chuzhou, southeastern Yangzhou, Nanjing, Zhenjiang, Liyang, Jintan, the boundary between Maanshan and Wuhu, and eastern Xuancheng. The regions with a moderate and high negative correlation (R ≤ −0.3) between NPP hum and population density accounted for 44.13% of the study area, mainly distributed in eastern and southern Huaian, western Yangzhou, southern Chuzhou, western Zhenjiang, Liyang, Maanshan, Wuhu, and eastern Xuancheng.
area. These regions were mainly located in western Xuancheng, Wuhu, Maanshan, yang, western Zhenjiang, Chuzhou, western Yangzhou, and the southern and north parts of Huaian. Among them, the regions with a highly significant negative correla (R ≤ −0.3, p < 0.01) between NPPhum and population density accounted for 35.48% of study area, mainly distributed in western Xuancheng, Wuhu, Maanshan, Liyang, wes Zhenjiang, southern Huaian, western Yangzhou, and Chuzhou. Figure 11. The spatial patterns of (a) the correlation coefficient between NPPhum and popula density, (b) and significant test of these data. Figure 11. The spatial patterns of (a) the correlation coefficient between NPP hum and population density, (b) and significant test of these data.
The regions with a significant positive correlation (R ≥ 0.3, p < 0.05) between NPP hum and population density accounted for 17.67% of the study area. These regions were mainly distributed in eastern Xuancheng, Nanjing, Jintan, central Zhenjiang, Chuzhou, and central Huaian. Among them, the regions with a highly significant positive correlation (R ≥ 0.3, p < 0.01) between NPP hum and population density accounted for 16.44% of the study area, and were mainly distributed in eastern Xuancheng, Jintan, central Zhenjiang, central Nanjing, and north-central Chuzhou. Figure 11b also shows that the regions with a significant negative correlation (R ≤ −0.3, p < 0.05) between NPP hum and population density accounted for 36.48% of the study area. These regions were mainly located in western Xuancheng, Wuhu, Maanshan, Liyang, western Zhenjiang, Chuzhou, western Yangzhou, and the southern and northern parts of Huaian. Among them, the regions with a highly significant negative correlation (R ≤ −0.3, p < 0.01) between NPP hum and population density accounted for 35.48% of the study area, mainly distributed in western Xuancheng, Wuhu, Maanshan, Liyang, western Zhenjiang, southern Huaian, western Yangzhou, and Chuzhou.

Conclusions
Based on the Thornthwaite Memorial model and residual analysis, this paper analyzed the spatial and temporal variations of vegetation NPP in the Nanjing metropolitan area and its response to climate change and human activities at the pixel scale. The results show that: (1) During the period 2000-2019, the NPP in the Nanjing metropolitan area showed a slow upward trend in general, and the NPP in the south of the Nanjing metropolitan area was higher than that in the north; (2) The NPP was influenced by both temperature and precipitation, and the impact of precipitation on NPP was greater than that of temperature. The increase in temperature and precipitation has led to an increasing trend of regional NPP; (3) Land use change significantly affected the regional NPP. The transformation of land use from low-NPP type to high-NPP type led to an increase in NPP, while the land use change from high-NPP type to low-NPP type was the main cause of regional NPP decline; (4) In the past 20 years, the NPP affected by human activities (NPP hum ) showed an upward trend, and human activities had a negative effect on NPP, which caused the actual NPP to be smaller than the potential NPP; (5) The NPP hum was influenced by both GDP and population density, and the impact of population density on NPP was greater than that of GDP. GDP was mainly positively related to NPP, while population density was mainly negatively correlated with NPP.