Effects of climatic factors on the net primary productivity in the source region of Yangtze River, China

The ecosystem of the Source Region of Yangtze River (SRYR) is highly susceptible to climate change. In this study, the spatial–temporal variation of NPP from 2000 to 2014 was analyzed, using outputs of Carnegie–Ames–Stanford Approach model. Then the correlation characteristics of NPP and climatic factors were evaluated. The results indicate that: (1) The average NPP in the SRYR is 100.0 gC/m2 from 2000 to 2014, and it shows an increasing trend from northwest to southeast. The responses of NPP to altitude varied among the regions with the altitude below 3500 m, between 3500 to 4500 m and above 4500 m, which could be attributed to the altitude associated variations of climatic factors and vegetation types; (2) The total NPP of SRYR increased by 0.18 TgC per year in the context of the warmer and wetter climate during 2000–2014. The NPP was significantly and positively correlated with annual temperature and precipitation at interannual time scales. Temperature in February, March, May and September make greater contribution to NPP than that in other months. And precipitation in July played a more crucial role in influencing NPP than that in other months; (3) Climatic factors caused the NPP to increase in most of the SRYR. Impacts of human activities were concentrated mainly in downstream region and is the primary reason for declines in NPP.


Scientific Reports
| (2021) 11:1376 | https://doi.org/10.1038/s41598-020-80494-9 www.nature.com/scientificreports/ middle and high latitudes while precipitation is the dominating factor in low latitudes 16 . However, some aspects of these relationships remain to be further studied and discussed, mainly due to the differences in study areas and study time periods. The Source Region of Yangtze River (SRYR) is a typical alpine region in the western Tibetan plateau, situated at 4000 m above mean sea level. It is referred to as an important ecological security shelter zone in Three-River Headwaters region. However, The SRYR's ecosystems and natural environment are inherently fragile and vulnerable to global warming. Similar to the whole Tibetan plateau, the SRYR has experienced significant warming trends in recent decades. The temperature increased by 0.34 °C/10a during the period from 1957 to 2013 17 and this warming is predicted to continue in the next 30 years based on CMIP5 Climate Models 18 . It is necessary to observe the vegetation and analyze the climate change in this significant place. This research took the SRYR as the study area and explored the effects of climatic factors on NPP. In this study, the hydrothermal climatic factors refer to precipitation and temperature, which can be obtained from meteorological data and the NPP data was simulated by CASA model.

Results
Spatial-temporal variation of climatic factors. The (Fig. 1b). From the perspective of the pattern of change, the multi-year average values of temperature and precipitation were generally lower in the western SRYR and increases gradually toward the east (Fig. 2a,b). During period 2000-2014, the positive changes of temperature and precipitation can be found in all parts of the SRYR. The increase trend grew more severe in the easterly direction and a remarkable increase occurred in middle and lower reaches of the SRYR (Fig. 2c,d).
Spatial-temporal variation of NPP. The spatial distribution of multi-year average NPP shows that NPP is gradually lower in the northwestern SRYR and increases gradually toward the southeast, with an average value of 100.0 gC/m 2 and a range from 0.2 to 260.9 gC/m 2 (Fig. 3a). The spatial correlation displayed in Fig. 3b showed that there may be an exponent relation between NPP and elevation. A dispersal NPP pattern along elevation ranging from 3500 to 4000 m could be detected. When the elevation is more than 4000 and less than 4500 m, NPP is sensitive to the change of elevation. NPP reduces by 23.9 gC/m 2 as the elevation increases by every 100 m. When the elevation is more than 4500 m, NPP is less susceptible to changes in elevation. The resulting slope indicates that a 100 m increase in elevation corresponds to a decrease in NPP by 8.4 gC/m 2 . Furthermore, the steppe is mainly located in the northwest, such as Tuotuo River Basin, Qumar River Basin and Beili River Basin, with lower NPP. However, the alpine meadow located in the southeast had higher NPP. These characteristics reflected the spatial heterogeneity of climate and terrain, and are in accordance with the gradients in humidity, temperature and elevation 19 .
The annual NPP in most SRYR (96.1% of total area) showed an increasing trend during the study period, with the change rate increasing from the west to the east. The most areas in the Dam River Basin and Middle stream experienced strongest positive trend (> 2.0 g C m −2 per year) (Fig. 4a). This result is consistent with the previous studies 11,20 . The change in annual NPP in the SRYR from 2000 to 2014 was calculated to determine the overall situation of NPP and the results are illustrated in Fig. 4b. It could be found that annual fluctuation in the total amount of NPP was not obvious during the study period, with a range of 12.       www.nature.com/scientificreports/ conclude that climate change is dominant factor impacting vegetation growth in the SRYR while the effect coming from human activities is much less. Climate change was responsible for NPP increase in 95.8% of the SRYR, whereas the human activities contributed to NPP decrease in 4.0% of the SRYR. The decreases in NPP caused by human activities were scattered, with some areas concentrated in Downstream Region. Combined impacts of climate change and human activities enhanced NPP in 0.2% of the SRYR.

Correlations between climatic factors and NPP. The spatial distribution of correlation coefficients
between climatic factors and NPP was presented by Figs. 6 and 7. It could be found that most areas of the SRYR showed a positive correlation between climatic factors and NPP. About 30.5% of the SRYR showed a significantly and positively correlation (p < 0.05) between annual temperature and annual NPP, primarily distributed in Qumar River Basin and Middle Stream Region (Fig. 6a). During April to September (growth season), significantly positive correlation (p < 0.05) between temperature and NPP was mostly observed in May (8.9%) and September (8.7%), namely the late spring and early autumn (Fig. 6c,g). Compared with temperature, the annual NPP was significantly and positively correlated with precipitation (p < 0.05) over more areas (40.4% of the SRYR), which were mainly found in Qumar River Basin, Middle Stream Region and Downstream Region (Fig. 7a). The significantly positive correlation between precipitation and NPP was more prevalent in July (32.1%), namely the middle summer (Fig. 7e).

Climatic controls on NPP in different vegetation types.
For the entire SRYR, annual NPP showed higher partial correlation with annual precipitation (R = 0.676, p < 0.01) than that with temperature (R = 0.618, p < 0.05). The responses of annual NPP to monthly climatic factors were varied. The annual NPP was significantly and positively correlated with temperature in September (R = 0.443, p < 0.1) and precipitation in July (p < 0.05). However, the partial correlations between NPP and climatic factors were insignificant in other months of growth season. For the three main vegetations, the NPP of meadow and steppe have similar responding characteristics    www.nature.com/scientificreports/ to climate factors. By the way, the NPP of meadow and steppe has higher sensitivity to annual mean precipitation than that to annual mean temperature. Besides, annual NPP of meadow and steppe also showed the stronger partial correlation with precipitation in July during the growth season. By contrast, the annual temperature played relatively more important roles in alpine vegetation. And NPP of alpine was correlated significantly with temperature in September during the growth season (Table 1).

Discussion
Comparison of the NPP simulation and change with previous studies. The 6 Although the methods and temporal scale used in these researches were different, due to the respective characteristics of study areas, the results of this research are reliable to some degree.
Variations in NPP along rising altitude. This research suggested that the multi-year average NPP is altitude-dependent. This characteristic is in accordance with previous studies 6,15 . When the elevation is lower than 4000 m, NPP increases as the elevation increases; when the elevation is higher than 4000 m, NPP decreases as the elevation decreases. The reason is that the area whose elevation is lower than 4000 m is mainly located in the downstream region, where the temperature will decrease as the elevation increases but the precipitation will increase as the elevation increases. What's more, from Figs. 6 and 7, it can be seen that the impact of precipitation to NPP is bigger than that of the temperature in the downstream region. Thus, in areas where the elevation is lower than 4000 m, as the elevation increases, the changing trend of NPP is similar with that of the precipitation (increasing trend). In areas where the elevation is higher than 4000 m, temperature and precipitation will both decrease with the increase of elevation, so the NPP will decrease. In addition, for the region located in 4000-4500 m, the sensitivity of NPP to elevation is higher than that in the region located above 4500 m. Previous research also indicated that the region located in 4000-4500 m is also the transition area from seasonally frozen ground region to permafrost region 25 . This maybe the reason why NPP has different sensitivities to elevation   Table 1. Partial correlation coefficients between NPP and climatic factors in SRYR. Note: *means the correlation is significant with p < 0.1; **means the correlation is significant with p < 0.05; ***means the correlation is significant with p < 0.01. www.nature.com/scientificreports/ proven that summer precipitation has a greater impact on vegetation growth in alpine regions 27,28 . This is mainly because the abundant precipitation will increase soil moisture and thus improve the availability of nutrients 29 . On the contrary, if precipitation does not increase correspondingly in the context of global warming, the vegetation growth will be inhibited 30 . Table 1 showed the relationship between annual temperature and monthly temperature within the growing season between NPP. It can be seen from Table 1 that the correlation between annual NPP and annual temperature is higher than that between annual NPP and monthly temperature within the growing season. It also can be concluded that temperature which is not in the growing season can influence annual NPP. Further analysis indicates that temperature of February and March have faire correlation with annual NPP (Fig. 8). The correlation analysis shows that when the mean monthly temperature of February and March increase by 1 °C, the total NPP of the SRYR will increase by about 0.43 TgC (R = 0.583, p < 0.05). Maybe it is caused by the temperature increase in the beginning of the growing season (the end of winter and the beginning of spring), which may help to stimulate photosynthetic enzyme activities from the cold environment and ignite vegetation growth through its impacts on nutrient availability and uptake [31][32][33] .
The annual NPP was highly and positively correlated with climatic variables in SRYR. We can conclude that climate change is dominant factor impacting vegetation growth in the SRYR while the effect coming from human activities is much less. The area where NPP decrease due to human activities are only 4.1% of the SRYR. Grazing is one of the primary factors to vegetation browning in this area, through vegetation cover reduction, top-layer soil degradation, soil compaction and so on 34 . In the Yushu Prefecture, mainly located in the SRYR, livestock numbers have decreased from 2.78 million in 2006 to 2.41 million in 2014 (Fig. 9a), due to grazing withdrawal www.nature.com/scientificreports/ policy. Although the overall grazing pressure reduced, the concentrations of grazing in specific regions (outside of the conservation area) became higher, which lead to regions of grassland degradation 22,35 . In addition, human populations have increased from 302.78 to 404.64 thousand during period 2006 to 2014 (Fig. 9b). With a rising population, the demand for material resources (e.g. milk, meat, fur products) increased continuously, which led to the livestock numbers cannot decreased significantly within a short time 36 .

Methodology
Study area. The Source Region of Yangtze River (SRYR for short, Latitude: 32° 25′ E and 35° 53′ E; Longitude: 89° 43′ E-97° 19′ E), located in the western Tibetan plateau, covers an area of 141,398 km 2 (Fig. 10a). The elevation ranges from 6456 m in the West to 3512 m in the East, with an average of 4779 m. The SRYR belongs to transition zone from semi-arid to semi-humid alpine area. The annual temperature is − 2 to − 3 °C. Monthly mean temperature in the coldest month is − 13.0 °C and that in the warmest month is 9.7 °C. The annual temperature of the study area is 265 mm. The temperature decreases from southeast to northwest 37 . The aridity index is 3.67 in the SRYR, which means the climate is very dry. The vegetation types are mainly meadow (84,985 km 2 ) and grassland (33,743 km 2 ), which are 60.1% and 23.9% (Fig. 10b)  The gridded data has been projected and resampled in order to ensure the same coordinate system and resolution with NDVI data. The NMIC also provides meteorological data of 9 meteorological stations within and around the study area, including parameters such as solar radiation, surface water, pressure, sunshine hours, wind speed and relative humidity. Grid data of the study area was interpolated by ANUSPLINE.
NPP simulation. In this study, the NPP were simulated by CASA (Carnegie-Ames-Stanford Approach) model. The CASA model is based on the plant growing mechanism 38-40 which can be summarized by Eq. (1).
where x and t are spatial location and time respectively, NPP is simulated value (gC m −2 ). APAR and ε represent absorbed photosynthetically active radiation and light use efficiency, which can be obtained by Eqs. (2) and (3).
where fPAR is the fraction of absorbed photosynthetically active radiation, SOL is the total solar radiation (MJ/ m 2 ), R is the fraction of solar active radiation that can be used by vegetation. T and W are temperature stress index and moisture stress factor, respectively. ε max is maximum light utilization efficiency. Further details of the above equations can be obtained from previous studies [38][39][40] .  www.nature.com/scientificreports/ The NPP calculated by CASA model can be considered as the actual NPP which is influenced by both climate change and human activities. It can be expressed as Eq. (4).
where PNPP and HNPP represent potential NPP and human-induced NPP, respectively. PNPP is only determined by climate conditions and without interference from human activities. It can be calculated by Thornthwaite Memorial model 41 , using the follows formulas: where t, L, N and v are average annual temperature (°C), annual maximum evapotranspiration (mm), annual total precipitation (mm) and average annual actual evapotranspiration (mm).
According to Eq. (4), the HNPP can be represented by the difference between PNPP and NPP.

Statistical analysis.
To identify the inter-annual trends of temperature (Tem.), precipitation (Pre.) and NPP, the linear regression method was adopted to eliminate the increase or decrease rate 42 , which can be calculated as follows: where θ slope is the linear slope of the time series variable, which can be used to characterize the increase or decrease rate during a given study period; n is the number of years (here n = 15); X i is the temperature, precipitation and NPP for the ith year (i = 1,2, … n).
A nonparametric test, Mann-Kendall (M-K) trend analysis 43,44 was utilized to detect the break points of temperature, precipitation and NPP series in the SRYR. The test statistic UF i is calculated as follows: where x i is the variable with the sample of n. E(S k ) and variance Var(S k ) could be estimated as follows: Using the same equation but in the reverse data series (x n , x n − 1 , …, x 1 ), UF i could be calculated again. Defining UB i = UF i (i = n, n − 1, …, 1), we can get the curve of UF i and UB i . If the intersection of the UF i and UB i curves occurs within the confidence interval, it indicates a change point 45 .
To assess the effects of temperature and precipitation on NPP in the SRYR, correlation coefficient R was employed to analyze the correlation between two variables (NPP vs. Tem., NPP vs. Pre.), using the following formula: where Y denotes the NPP and X denotes temperature or precipitation.
The results of the statistical analysis above can be got by MATLAB.

Identification of the relative roles of climate change and human activities in NPP. A positive
PNPP slope indicates that vegetation growth is promoted by climate change, whereas a negative PNPP slope means that climate change reduced the vegetation NPP. A positive HNPP slope suggests that human activities 1 + 1.05N L 2 (7) L = 300 + 25t + 0.05t 3 Var(S i ) (i = 1, 2, . . . , n) (10) S k = k i=1 r i (k = 2, 3, . . . , n) (11) ri = +1 x i > x j 0 x i ≤ x j (j = 1, 2, . . . , i − 1) www.nature.com/scientificreports/ have negative influence on vegetation growth and create ecological degradation, whereas a negative HNPP slope means that human activities contribute to vegetation growth 46 . Thus, the determinants for NPP change can be identified according to Table 2.