Impact of rural depopulation and climate change on vegetation, runoff and sediment load in the Gan River basin, China

Climate change and rural depopulation are changing the ecological and hydrological cycles in China. Data on the normalized difference vegetation index (NDVI), temperature, precipitation, stream ﬂ ow, sediment and rural population are available for the Gan River basin from 1981 to 2017. We investigated the spatio-temporal variations in climate, human activity and vegetation mainly using the Mann – Kendall test and examined their relationship using the Granger causality test. The results showed that (1) the temperature markedly increased in all seasons; (2) the precipitation increased in summer and winter but decreased in spring and autumn; (3) overall, the NDVI increased markedly during 2005 – 2017, but showed seasonal differences, with decreases in summer and winter and increases in spring and autumn; (4) the annual sediment transport showed a signi ﬁ cant decreasing trend and (5) a large number of the population shifted from rural to urban areas, resulting in a decrease in the rural population between 1998 and 2018. Rural depopulation has brought about farmland abandonment, conversion of farmland to forests, which was the factor driving the recovery of the vegetation and the decrease in sediment. The results of this study can provide support for climate change adaptation and sustainable development.


INTRODUCTION
According to the Intergovernmental Panel on Climate Change (IPCC) Fifth Assessment Report (AR5), global climate change is unquestionable. The urbanization process has also been developing rapidly over the past 30 years in China. Urbanization is a dynamic process generated by the population exchange from rural to urban areas through internal migration (Ledent & Jacques ). The United Nations predicts that population growth of urban areas will represent the bulk of the world's projected population growth, rising to 6.3 billion by 2050 (While & Whitehead ). Additionally, some developed countries in Asia will In order to quantify the specific causal relationship between eco-hydrological elements and human activities and climate change, the Granger causality test (GCT) can be used as a reference method. The GCT (Granger ), which was first proposed in 1969 by Clive W. Granger, the Nobel laureate in economics, defines causality as 'the ability of a priori value of one time series to predict another time series' when predicting the future of a set of data. If predicting future information using the past information of another set of data is better than using only that data itself, then there is a Granger causality between the two sets of data (Papagiannopoulou et al. ). In China, how do ecological and hydrological conditions change in the context of climate change and rural depopulation? What impact do climate change and rural depopulation have on the eco-hydrological process? In this study, data for precipitation, temperature, population, normalized difference vegetation index (NDVI), runoff and sediment load were available for the Gan River basin from 1981 to 2017. The trend and relationship of eco-hydrological factors will be analyzed to determine the impact of rural depopulation and climate change on the ecological and hydrological processes by using the Mann-Kendall (M-K) test and the GCT.
The organization of the paper is as follows: The next section describes the study area and data; the following two sections introduce the methodologies and present the results and discussion; and the final section provides the conclusions.
The Gan River is a tributary of the Yangtze River, with a drainage area of 82,809 km 2 , occupying more than 50% of Jiangxi Province (Zhang et al. a, b). The Gan River basin is in a humid subtropical monsoon climate zone with abundant precipitation and frequent heavy rain. The mean annual precipitation is about 1,505 mm , and the mean For analyzing the long-term dynamics of high-quality data, linear regression was applied to monthly MODIS NDVI and GIMMS NDVI data that were averaged over the entire study area. When using the temporal overlap (2000.2-2006.12) of monthly GIMMS and MODIS, the data fitting was reliable (R 2 > 0.7), as shown in Figure

METHODOLOGIES M-K test
In this paper, the M-K test was used to detect the trends in seasonal temperature, precipitation and NDVI. The M-K test is a nonparametric testing method (Mann ; Kendall ). Samples are not required to follow a certain distribution, nor are they interfered with by a few outliers. It is recommended and widely used by the World Meteorological Organization. Hence, it is suitable for non-normal distribution data in hydro-meteorological time series.
The statistic S is calculated using the following equations: where {x 1 , x 2 , . . . , x n } is the time-series data, and n is the number of data points. E(S) and var(S) are the mean and variance of the statistic S, respectively. When n > 10, the

Granger causality test (GCT)
In this paper, the GCT was mainly used to examine the causal relationships among the climatic factors, vegetation, water and sediment variables and socioeconomic factors.
The steps of the GCT are as follows: Step Step 2: Select a suitable lag term m, set up the auto-regressive model and calculate the residual sum of squares (RSS R ): where the term y t is the dependent variable at time t, a 0 is a constant term, a i is the regression coefficient, and ε t is the error term.
Step 3: Select the appropriate lag value q for time series X to establish the regression and calculate the residual sum of squares (RSS UR ): Step 4: Assume that H 0 : items of all time series do not belong to the regression.
Step 5: Perform the F-test: where n is the sample size.

RESULTS AND DISCUSSION
Trend analysis of annual temperature, precipitation and NDVI The Gan River basin was divided into 329 sub-basins based on the DEM data. All observed temperature, precipitation and NDVI data were interpolated to each sub-basin by the IDW method (inverse distance weighting method) (Shepard ).        To further quantitatively analyze the relationship between rural depopulation/climate change and eco-hydrological processes, the GCT was carried out on an annual scale on the precipitation and temperature, NDVI, socioeconomic factors, runoff and sediment variables in the entire Gan River basin. It was found that the influence of the precipitation and temperature on the NDVI was not significant on the annual scale. We can see from Figure 9 that the rural population (RP) had a significant effect on the NDVI changes (p ¼ 0.08).
The effect of precipitation on the streamflow was significant (p < 0.1). The influence of the NDVI on streamflow on the annual scale was not significant (p ¼ 0.46). In addition, the effect of the NDVI on sediment transport (p ¼ 0.15) was relatively obvious compared with the effect of the average flow on sediment transport (p ¼ 0.58) on the annual scale. which regulated the amount of sediment to a certain degree.

CONCLUSIONS
In the context of climate change and urbanization, the temperature of the Gan River basin has shown a significant increase.
Precipitation has increased in spring and autumn, and decreased in summer and winter. Its vegetation has been continuously restored, and runoff has not changed significantly, but sediment load has shown the downward trend, the rural population showed a decreasing trend. Here, the GCT was used to analyze the statistical causality of these variables. The effect of climate change on the vegetation-hydrological process was found to be not significant on the annual scale, but rather one of the main reasons for the vegetation restoration and sediment load reduction was rural depopulation.
This study can provide reference for studying the rural depopulation effect in other areas of the world, such as India and Vietnam that are rapidly urbanizing. However, a comprehensive analysis of the reasons for the ecological and hydrological changes in the Gan River basin requires further studies in the future.