**4. Discussion**

Long et al. [10] used the normalized difference vegetation index (NDVI) to study the response of vegetation to climatic factors in IMAR from 1982 to 2006 and found that different vegetation types had different responses to climatic factors and different time lag, but only air temperature and precipitation were taken into consideration. Based on the MODIS data, Mu, et al. [11] studied the response of vegetation cover to air temperature and precipitation in IMAR from 2001 to 2010. It was suggested that the relationship between vegetation cover and precipitation in IMAR was closer, and there was a certain lag in the response of vegetation growth and climatic factors but the exact time lag was not found. The NDVI data of vegetation was used to study the lagged response of precipitation in Xilingol League of IMAR from 2001 to 2007. Liu et al. [52] believed that the vegetation growth of Xilingol League in the growing season has an obvious lag to precipitation, and the time lag was about 50–60 days. Different types of steppe have different time lag. Li et al. [32] used the field sampling data and climate data from the Inner Mongolia Steppe Ecosystem Research Station of the Chinese Academy of Sciences to study the response of Carex korshinskyi to climatic factors in 2014 and found that the biomass of Carex korshinskyi lagged behind air temperature by 16 days and the relative humidity by 15 days.

The zoning of driving force in this paper shows that precipitation driving areas (21.8%) in IMAR were much larger than the air temperature driving areas (8%) and relative humidity driving areas (11.6%), which was basically consistent with the view of Mu et al. [11]. Since this paper was based on the result of time lag analysis, the response of the forest ecological areas to precipitation and relative humidity was generally higher than that of air temperature in the spatial distribution of different zones, and the response of the steppe ecological areas to precipitation and air temperature was generally higher than that of relative humidity. Regarding the time lag of EVI in the growing season, we draw a conclusion that the time lag of EVI in IMAR to different climatic factors was about 1–2 months for air temperature, 1–2 months for relative humidity, and one month for precipitation. On one hand, data sources (especially time resolution), the selection of analysis index, and experimental errors may cause differences in results. On the other hand, the spatial resolution of remote sensing data, the size

of interpolation grid for climatic factors, and the changes in ecological complexity may also lead to large differences in the response and time lag of the EVI to climatic factors in IMAR [11,32,52].

This paper mainly discusses the relationship between changes of IMAR vegetation EVI and climatic factors (air temperature, relative humidity, and precipitation). In addition, there was also a certain correlation between vegetation EVI and extreme weather [15]. In the desert biota of IMAR, other climate factors like wind speed also had a great impact on local biomass [54]. Changes in vegetation EVI were not only closely related to climate factors but also related to human activities and other factors [15,55,56]. Before the 21st century, IMAR had a declining trend in overall vegetation growth under the combined effects of climate change, social economy, and local policy changes [57,58]. Since the beginning of the 21st century, China has implemented a series of ecological restoration projects, such as harnessing wind and sand in the Beijing-Tianjin region, returning farmland to forests and grasslands, returning grazing land to grassland, enclosing and transferring, and IMAR is one of the key implementation areas for these projects [11,59]. At the beginning of the study period, IMAR has begun implementing state-owned afforestation, encouraging artificial afforestation, returning farmland to forests and grasslands, closing mountains for afforestation, and new closures of non-forested land and open forest land, which has resulted in effective protection and improvement of forest lands and grasslands, especially forest lands [11,26,52]. In semi-arid areas, human management of ecological zones has become a major driving force of vegetation change [54]. Similar to many studies, this study find that the vegetation in some dusty land and desert steppe (such as SonidLeft Banner and Etuoke Banner) in IMAR remained a good growing trend from 2000 to 2015, and the implementation of ecological restoration measures (such as planting tree and grass, prohibiting grazing, preventing and governing the sand) played an important role in this process [10,11,59]. According to the IMAR Statistical Yearbook [60], from 2000 to 2002, the crop area affected by natural disasters in the IMAR reduced from 48,000 km<sup>2</sup> to 32,000 km2, the total number of grazing livestock decreased from 73.01 million to 63.27 million, the area of arable land reduced from 73,000 km<sup>2</sup> to 69,000 km2, the area of artificial afforestation increased from 5900 km<sup>2</sup> to 9100 km2, and the population decreased from 23.77 million to 22.79 million. The EVI index increased significantly from 0.251 in 2001 to 0.278 in 2002 under the combined effect of reduced intensity in natural disasters, grazing and human activity, in addition to greater intensity in returning farmland to forests and grasslands, and the impact of climate change. Subsequently, from 2002 to 2007, the area of artificial afforestation decreased from 9100 km<sup>2</sup> to 5900 km2, the total population increased from 22.79 million to 24.05 million, the total number of livestock increased from 63.27 million to a maximum of 100.04 million, resulting in an increase of grazing intensity. The superposition effect resulted in a slow decline in the EVI index between 2002 and 2007, with the lowest mean value of 0.259. From 2007 to 2008, the total number of livestock decreased by 5% and the area of artificial afforestation increased to 7200 km2, with EVI reaching a peak mean value of 0.281 in 2008. From 2008 to 2011, the total population, the crop area affected by natural disasters and the total number of livestock increased slowly, while the area of artificial afforestation remains basically unchanged, therefore, the EVI of IMAR vegetation slowly declined from 2008 to 2011. In 2012, the area of artificial afforestation increased to 7800 km2, and the crop area affected by natural disasters fell to 26,000 km2. Therefore, the EVI of IMAR increased significantly by 10.44%, reaching the highest mean value of 0.301 during the study period. After 2012, the area of artificial afforestation remained basically unchanged, but the total number of livestock increased by 20%. The crop area affected by natural disasters increased significantly from 26,000 km2 in 2012 to 101,000 km2 in 2015. This, together with the impact of climate change, led to a declining trend of EVI in IMAR from 2012 to 2015. In general, from 2000 to 2015, the total population increased by 6.3%, the total number of livestock increased by 65.7%, so the increase in the intensity of human activities year by year will cause the growth rate of vegetation in IMAR to slow down. Human disturbance at different time periods has different effects on vegetation growth, and such positive or negative influences will increase the inter-annualecological complexity of IMAR, and change the response

sensitivity of vegetation EVI to climate factors. Eventually, it led to differences in the response degree and response speed of EVI to air temperature, relative humidity, and precipitation [11].
