Climate change drives NDVI variations at multiple spatiotemporal levels rather than human disturbance in Northwest China.

Changes in land management and climate alter vegetation dynamics; however, the factors driving vegetation changes remain elusive at multiple spatiotemporal levels. Here, we assess the drivers of changes in greenness from 2000 to 2015 in Northwest China (NW China). We used multiple stepwise linear regression (MSLR), redundancy analysis (RDA), and 12 other models to quantify the impacts of precipitation and temperature metrics, gross domestic product (GDP), population, and grazing intensity on the normalized difference vegetation index (NDVI) at three administrative levels (county, town, and village), four temporal levels (yearly, May, July, and September), two vegetation types (woodland and grassland), and at annual precipitation gradients of <200, 200-400, and >400 mm. The results suggest that NW China underwent vegetation greening from 2000 to 2015. Precipitation and temperature were the most influential factors contributing to the NDVI change. Population was the main determinant of NDVI under the precipitation gradient of <200 mm, and the effect of GDP on NDVI was moderate. On the temporal scale, annual precipitation, precipitation before the previous year, and precipitation in the current year determined the NDVI in May, July, and September, respectively, for both woodland and grassland. At multiple scales, climate change was the primary driver of vegetation change in NW China, rather than human disturbance. These findings expand our understanding on drivers of NDVI at multiple levels over a long period. Measures to manage decreasing vegetation coverage may be more effective and could be implemented sooner based on predicted climate change in drylands worldwide.