EN中文

Journal of Geographical Sciences >

2021 , Vol. 31 >Issue 6: 785 - 801

DOI: https://doi.org/10.1007/s11442-021-1871-7

Research Articles

Sensitivity and areal differentiation of vegetation responses to hydrothermal dynamics on the northern and southern slopes of the Qinling Mountains in Shaanxi province

  • QI Guizeng , 1, 2, 3 ,
  • BAI Hongying , 1, 2, 3, * ,
  • ZHAO Ting 1, 2, 3 ,
  • MENG Qing 1, 2, 3 ,
  • ZHANG Shanhong 1, 2, 3
Expand
  • 1. Shaanxi Key Laboratory of Earth Surface System and Environmental Carrying Capacity, Northwest University, Xi'an 710127, China
  • 2. Institute of Qinling Mountains, Northwest University, Xi’an 710127, China
  • 3. College of Urban and Environmental Science, Northwest University, Xi’an 710127, China
* Bai Hongying (1962–), Professor, specialized in global change ecology and physical geography. E-mail:

Qi Guizeng (1994-), Master Candidate; specialized in regional climate change and vegetation response. E-mail:

Received date: 2021-01-20

  Accepted date: 2021-03-28

  Online published: 2021-08-25

Supported by

Key Research and Development Program of Shaanxi Province(2019ZDLSF05-02)

Key Research and Development Program of Shaanxi Province(2020SF-400)

Shaanxi Province Water Conservancy Science and Technology Project(2020slkj-13)

Fold

Abstract

The Qinling Mountains, located at the junction of warm temperate and subtropical zones, serve as the boundary between north and south China. Exploring the sensitivity of the response of vegetation there to hydrothermal dynamics elucidates the dynamics and mechanisms of the main vegetation types in the context of changes in temperature and moisture. Importance should be attached to changes in vegetation in different climate zones. To reveal the sensitivity and areal differentiation of vegetation responses to hydrothermal dynamics, the spatio-temporal variation characteristics of the normalized vegetation index (NDVI) and the standardized precipitation evapotranspiration index (SPEI) on the northern and southern slopes of the Qinling Mountains from 2000 to 2018 are explored using the meteorological data of 32 meteorological stations and the MODIS NDVI datasets. The results show that: 1) The overall vegetation coverage of the Qinling Mountains improved significantly from 2000 to 2018. The NDVI rise rate and area ratio on the southern slope were higher than those on the northern slope, and the vegetation on the southern slope improved more than that on the northern slope. The Qinling Mountains showed an insignificant humidification trend. The humidification rate and humidification area of the northern slope were greater than those on the southern slope. 2) Vegetation on the northern slope of the Qinling Mountains was more sensitive to hydrothermal dynamics than that on the southern slope. Vegetation was most sensitive to hydrothermal dynamics from March to June on the northern slope, and from March to May (spring) on the southern slope. The vegetation on the northern and southern slopes was mainly affected by hydrothermal dynamics on a scale of 3-7 months, responding weakly to hydrothermal dynamics on a scale of 11-12 months. 3) Some 90.34% of NDVI and SPEI was positively correlated in the Qinling Mountains. Spring humidification in most parts of the study area promoted the growth of vegetation all the year round. The sensitivity of vegetation responses to hydrothermal dynamics with increasing altitude increased first and then decreased. Elevations of 800 to 1200 m were the most sensitive range for vegetation response to hydrothermal dynamics. The sensitivity of the vegetation response at elevations of 1200-3000 m decreased with increasing altitude. As regards to vegetation type, grass was most sensitive to hydrothermal dynamics on both the northern and southern slopes of the Qinling Mountains; but most other vegetation types on the northern slope were more sensitive to hydrothermal dynamics than those on the southern slope.

Key words： Qinling Mountains in Shaanxi province; NDVI; SPEI; sensitivity; spatial difference

Cite this article

QI Guizeng , BAI Hongying , ZHAO Ting , MENG Qing , ZHANG Shanhong . Sensitivity and areal differentiation of vegetation responses to hydrothermal dynamics on the northern and southern slopes of the Qinling Mountains in Shaanxi province[J]. Journal of Geographical Sciences, 2021 , 31(6) : 785 -801 . DOI: 10.1007/s11442-021-1871-7

1 Introduction

Vegetation is an important component of terrestrial ecosystems, and it plays an important role in the mutual regulation of atmosphere and soil and water, and in the optimization of ecosystem services (Fu et al., 2017; Gao et al., 2019). Vegetation and climate change are inextricably linked, and vegetation dynamics not only reflects trends in climate change, but actually has some adaptability to climate change (Liu et al., 2016), making it a good indicator of such change (Thuiller, 2005). The normalized vegetation index (NDVI) is a commonly used index that reflects the growth of vegetation, and one that is closely related to the primary productivity of vegetation and leaf area index, accurately reflecting vegetation cover and growth (Hou et al., 2013). In the context of continuous global warming, the warming leads to an acceleration of the water cycle and a change in water balance processes, such as evaporation from the surface and transpiration by plants (Stocker, 2013), as well as a significant upward trend in the frequency, severity, and duration of droughts (Dai, 2012; Huang et al., 2016). The impact of hydrothermal dynamics on vegetation growth is becoming more and more evident and has become a hot issue in today’s research. The standardized precipitation evapotranspiration index (SPEI) (Vicenteserranoet al., 2010a) is commonly used to characterize hydrothermal dynamics conditions and has the same sensitivity to temperature as the Palmer drought severity index (PDSI) (Palmer, 1968). The SPEI has the advantages of being sensitive to the temperature and standardized precipitation index (SPI) (Mckee et al., 1993) and can be used in many areas to assess the characteristics of regional hydrothermal dynamics (Byakatonda et al., 2018; Tong et al., 2018; Qi et al., 2019). Therefore, the SPEI was used in combination with the NDVI to investigate the sensitivity of the vegetation response to hydrothermal dynamics.
Many scholars have studied the relationship between vegetation and climatic factors in different regions, finding that moisture is a significant climatic factor controlling NDVI in northern China and the Qinghai-Tibet Plateau; while temperature plays a dominant role in eastern, central, and southwestern China, where accelerated warming can have a suppressive effect on vegetation growth through mechanisms such as increased drought (Gao et al., 2019). The rise in drought indices, such as SPEI, has a catalytic effect on NDVI in most regions of China, especially in mid- and high-altitude regions, but NDVI is less affected by SPEI in regions with relatively abundant precipitation, such as the southeastern Yangtze River basin and the lower Pearl River basin, and in alpine regions, such as northeastern Heilongjiang (Kong et al., 2016). Vegetation in grassland and mid-to-high-elevation regions in north China is sensitive to the SPEI response (Yang et al., 2018) and is most affected by hydrothermal dynamics in summer (Yang et al., 2018). On the northern slope of the Tianshan Mountains, vegetation is most affected by hydrothermal dynamics in spring and summer (Li et al., 2019). The above studies indicate that there are significant spatio-temporal differences in the vegetation response to climate change in different regions, but the mechanisms of the response in different climatic conditions in the north-south transitional zone of China are still unclear.
The Qinling Mountains (QMs) are at the junction of a temperate monsoon climate and subtropical monsoon climate, and are also an important north-south geographic boundary in China. The QMs exhibit significant variation in their north-south climate zones and vegetation zones (Kang and Zhu, 2007), making them an ideal area in which to study the response patterns of vegetation to climate change in different climate zones from a small regional perspective. In this paper, based on the MODIS NDVI remote sensing dataset and the meteorological data from 32 meteorological stations for the years 2000-2018, we explored the spatio-temporal changes in the NDVI and SPEI on the northern slope of the Qinling Mountains (NSQM) and southern slope of the Qinling Mountains (SSQM). The spatio-temporal variations of NDVI and SPEI and the sensitivity of the vegetation response to hydrothermal dynamics and spatial differences reflect patterns in and the mechanical characteristics of major vegetation types in the context of their response to hydrothermal dynamics in subtropical and warm temperate regions; they also provide a scientific basis for the construction of ecological civilization in the QMs and the study of ecosystem responses to climate fluctuations.

2 Materials and methods

2.1 Study area and data sources

The QMs in this paper refer to the Qinling Mountains in a narrow sense (i.e., the Qinling Mountains in Shaanxi Province), between 32°40°N-34°35°N and 105°30°E-111°03°E, with a total area of 61,900 km2 and an elevation of 195-3767.2 m. The QMs exhibit a sensitive response to climate change (Kang and Zhu, 2007), and the geographic boundary between north and south China is generally coincident with the annual 800 mm isohyet line, the 0℃ isotherm in January, and the 2000 h sunshine hour line. The NSQM is steep and has a warm-temperate, semi-humid climate, while the SSQM is gentle and long and has a northern subtropical humid climate (Deng et al., 2018b; Zhang et al., 2018; Lu and Lu, 2019).
The meteorological data were sourced from the Shaanxi Meteorological Bureau and China Meteorological Data Network (http://data.cma.cn/) and were checked for consistency and quality control. The data include monthly mean temperature and precipitation data from 32 meteorological stations on the NSQM and SSQM from 2000 to 2018 and meteorological data from 11 meteorological stations at high altitude in the Taibai Mountains from 2013 to 2015 (for spatial interpolation). The distribution of meteorological stations is shown in Figure 1.
Figure 1 Geographical environment and distribution of meteorological stations on the northern and southern slopes of the Qinling Mountains
NDVIs during 2000-2018 were obtained from the MODIS/Terra Vegetation Indices 16-Day L3 Global 250 m SIN Grid (MOD13Q1) product of NASA. The NDVI dataset is spatially resolved at 250 m×250 m and temporally resolved at 16 days. The monthly NDVI data are synthesized using MODIS Reprojection Tool (MRT) for image stitching, projection, and format conversion and a maximum value composite (MVC) to remove the effects of clouds, atmosphere, and solar altitude angle (Denget al., 2018a). The vegetation cover type data were sourced from the Resource and Environmental Science Data Center of the Chinese Academy of Sciences (http://www.resdc.cn) with a resolution of 1:100,000.

2.2 Research methodology

In this paper, a change in hydrothermal dynamics at multiple timescales is expressed as follows: SPEI-1 (a timescale of 1 month) is the monthly hydrothermal dynamics level; SPEI-3 is the seasonal hydrothermal dynamics level; SPEI-4 is the timescale of a four-month hydrothermal dynamics level, and SPEI-12 is the timescale of an annual hydrothermal dynamics level (Vicenteserrano et al., 2010b). For example, SPEI-3 in May indicates the combined hydrothermal dynamics level of spring during March-May, and SPEI-4 in June indicates the combined hydrothermal dynamics level during March-June.
The spatial distribution of the SPEI index was obtained by interpolation using the ANUSPLIN method (Hutchinson and Xu, 2013). The ANUSPLIN method is suitable for the interpolation of meteorological elements (Liu et al., 2008), and the interpolation accuracy error is small for a complex mountain environment (Xu et al., 2018). At present, this method is the best interpolation method for the QMs (Meng et al., 2019) and has been used extensively for the spatial interpolation of SPEIs (Dong et al., 2013). Therefore, this paper uses the ANUSPLIN method for the spatial interpolation of SPEIs in the QMs. Based on the climatic characteristics of the QMs in this study, spring is from March to May, summer is from June to August, autumn is from September to November, and winter is from December to the following February.
The spatial interpolation accuracy was examined using the mean absolute error (MAE) and root mean square error (RMSE) of the SPEI values of 11 meteorological stations in the Taibai Mountains from 2013 to 2015. The smaller the MAE and RMSE values, the better the spatial interpolation results (Bai et al., 2012). The results showed that the spatial interpolation accuracy of the annual SPEI and spring SPEI was high and met the requirements of this study (Table 1).
Table 1 Results of SPEI spatial interpolation on the northern and southern slopes of the Qinling Mountains
Year MAE RMSE
Yearly interpolation test Spring interpolation test Yearly interpolation test Spring interpolation test
2013 0.137 0.140 0.171 0.173
2014 0.043 0.062 0.064 0.093
2015 0.240 0.343 0.318 0.449
The tendency rates of the NDVI and SPEI were calculated using the linear regression of one-variable trend analysis method and analyzed for significance; the tendency rates were calculated as follows:
$slope=\frac{n\mathop{\sum }_{i=1}^{n}i{{x}_{i}}-\mathop{\sum }_{i=1}^{n}i\mathop{\sum }_{i=1}^{n}{{x}_{i}}}{n\mathop{\sum }_{i=1}^{n}{{i}^{2}}-{{\left( \mathop{\sum }_{i=1}^{n}i \right)}^{2}}}$
where n is the time series; xi is the value of year i (NDVI or SPEI index); slope>0 orslope<0 indicates an increasing or decreasing trend inx.
Whether there is a correlation between two variables or between the magnitude of a correlation based on the site data and an image element scale is determined by the following formula:
${{R}_{xy}}=\frac{\mathop{\sum }_{i=1}^{n}\left[ \left( {{x}_{i}}-\bar{x} \right)\left( {{y}_{i}}-\bar{y} \right) \right]}{\sqrt{\mathop{\sum }_{i=1}^{n}{{\left( {{x}_{i}}-\bar{x} \right)}^{2}}\mathop{\sum }_{i=1}^{n}{{\left( {{y}_{i}}-\bar{y} \right)}^{2}}}}$
where Rxy is the correlation coefficient between variable SPEI and variable NDVI, xi is the value of variable SPEI in year i, yi is the value of variable NDVI in year i, and $\bar{x}$ and $\bar{y}$ are the mean values of variables SPEI and NDVI, respectively.
Significance of trend and correlation tests were determined using t-test, and the results were classified as extremely significant (p≤0.01), significant (0.01<p≤0.05), weakly significant (0.05<p≤0.1), and insignificant (p>0.1).

3 Results

3.1 Spatial variation characteristics of annual NDVI and annual SPEI on the NSQM and SSQM

Figure 2 shows the spatial distribution of the annual change rate of NDVI and significance of the NSQM and SSQM during 2000-2018. Figure 2a shows the annual change rate of NDVI in the QMs from 2000 to 2018, ranging from -0.37 to 0.21/10a: 90.88% of the areas in the QMs show an increasing trend, and only 9.12% of the areas in the QMs show a decreasing trend (with low elevation and gentle terrain on both sides of the NSQM and SSQM; with Chang’an and Huxian on the NSQM and Hanzhong and Ankang on the SSQM as the representative areas). The significance test (Figure 2b) shows that 72.36% of the areas in the QMs have a significant increasing trend in NDVI, indicating that vegetation cover in the QMs improved significantly.
Figure 2 Spatial distribution of the trend and significance of the NDVI on the northern and southern slopes of the Qinling Mountains during 2000-2018
The average annual change rate of NDVI on the NSQM was 0.017/10a: 79.31% of the areas on the NSQM show an increasing trend, and the percentage of areas with a significant increase was 53.15%; the average annual change rate of NDVI on the SSQM was 0.031/10a: 93.92% of the areas on the SSQM show an increasing trend, and the percentage of areas with a significant increase was 77.39%. From 2000 to 2018, the increase in NDVI on the SSQM was higher than that on the NSQM, and the improvement in vegetation on the SSQM was more marked than that on the NSQM.
Figure 3 shows the spatial distribution of the annual SPEI tendency rate and significance of the NSQM and SSQM from 2000 to 2018. Figure 3a shows the annual SPEI tendency rate of the QMs, ranging from -0.54 to 0.77/10a: 74.68% of the areas in the QMs show an increasing trend, and 25.32% of the areas in the QMs show a decreasing trend. This indicates that the wet area is much larger than the arid area of the QMs from 2000 to 2018—the wetted area of Qinling is much larger than the arid land area of Qinling.
Figure 3 Spatial distribution of the tendency rate and significance of SPEI on the northern and southern slopes of the Qinling Mountains during 2000-2018
The spatial distribution of significance (Figure 3b) shows that 96.63% of the areas in the QMs have insignificant changes in SPEI. The average annual change rate of SPEI on the NSQM is 0.13/10a: 87.18% of the areas on the NSQM have an increasing trend in SPEI. The average annual change rate of SPEI on the SSQM is 0.022/10a: 71.43% of the areas on the SSQM have an increasing trend in SPEI. The above indicates that the overall hydrothermal dynamics of the QMs are relatively stable, and that the wetting trend is not significant; but the wetting rate and percentage of the wetted area of the NSQM are larger than those of the SSQM.

3.2 Differences in the sensitivity of the vegetation response to multi-scale hydrothermal dynamics on the NSQM and SSQM

Figure 4 shows plane and stereo plots of annual NDVI and multiple timescales of SPEI correlation coefficients for the NSQM and SSQM for 2000-2018. Figures 4a and 4b show a clear regularity in the response of vegetation on the NSQM and SSQM to hydrothermal dynamics at multiple timescales. The positive correlations on the NSQM and SSQM are mainly concentrated in March-July, and the negative correlations, mainly in January, February, and August-December. The positive correlations reach significance mainly at timescales of 3-7 months in May and June. The correlations change from negative to positive and then to negative from January to December; and the correlations weaken with increasing timescale of SPEI. This indicates that vegetation growth on the NSQM and SSQM is more affected by shorter timescales (3-7 months) of hydrothermal dynamics in May and June, and less affected by longer timescales of hydrothermal dynamics (11-12 months).
Figure 4 Correlation coefficients between annual NDVI and SPEI of each month
A comparison of Figures 4a and 4b shows differences in the sensitivity of the vegetation response to hydrothermal dynamics between the NSQM and SSQM. The response of annual NDVI to hydrothermal dynamics is significant or extremely significant in more months and timescales on the NSQM than on the SSQM. The significant or extremely significant correlations between NDVI and multi-scale SPEI in each month are mainly concentrated in different timescales in June on the NSQM, and mainly in different timescales in May on the SSQM. The above indicates that vegetation on the NSQM is more sensitive than vegetation on the SSQM to multiple-timescale hydrothermal dynamics in different months; and that vegetation on the NSQM is mainly affected by multiple timescales of hydrothermal dynamics in June, while vegetation on the SSQM is mainly affected, severely, by multiple timescales of hydrothermal dynamics in May.
From the timescale of a single month of SPEI, the first columns of Figures 4a and 4b show the annual NDVI and single-month SPEI (SPEI-1) correlation coefficients for the NSQM and SSQM, respectively. The annual NDVI of the NSQM is significantly positively correlated with SPEI-1 in March, April, May, and June, respectively. The correlation between SPEI-1 and the annual NDVI reaches significance in March, but becomes negative from July onwards. The annual NDVI on the southern slope is significantly positively correlated with SPEI-1 in March, April, and May, respectively; and the correlation between SPEI-1 and the annual NDVI reaches significance in May and becomes negative in June. Therefore, on a monthly scale, the vegetation responses to hydrothermal dynamics on the NSQM and SSQM differ: on the NSQM, wetting in March significantly promotes vegetation growth, and wetting from July onwards suppresses vegetation growth; on the SSQM, wetting in May significantly promotes vegetation growth, and wetting from June onwards suppresses vegetation growth.
For timescales of multiple months of SPEI (columns 2-12 of Figures 4a and 4b), the largest and extremely significant positive correlation (r=0.68, p≤0.01) was found between the annual NDVI and SPEI-4 in June (combined hydrothermal dynamics levels from March to June) for the NSQM. The annual NDVI was also extremely significantly positively correlated with both SPEI-5 and SPEI-6 in June, but the correlation coefficients were reduced. Because SPEI-5 in June (combined hydrothermal dynamics levels from February to June) included SPEI-4 in June (combined hydrothermal dynamics from March to June), which showed an extremely significant positive correlation. The highest and extremely significant positive correlation coefficient (r = 0.65, p≤0.01) was found between the annual NDVI and SPEI-3 in May (SPEI of spring; combined hydrothermal dynamics levels from March to May), while SPEI-4, SPEI-5, SPEI-6, and SPEI-7 in May also showed extremely significant positive correlations due to their inclusion SPEI-3 in May.
The above indicates that vegetation growth on the NSQM is most sensitive to the overall hydrothermal dynamics from March to June, and that vegetation growth on the SSQM is most sensitive to the hydrothermal dynamics in spring (March-May). This conclusion is consistent with the conclusion of the tree chronologic study: that climate factors from March to June are most closely related to the radial growth of natural forests in the QMs and have a significant influence on the annual Net Primary Productivity (NPP) in the QMs.
In order to reveal more comprehensively the relationship between vegetation growth and multiple timescales of hydrothermal dynamics for each month on the NSQM and SSQM, correlation coefficient plots were constructed in three dimensions. Figures 4c and 4d show that the annual NDVI of the NSQM and SSQM are positively correlated with the multiple timescales of SPEI in March-July, and negatively correlated with the multiple timescales of SPEI in January-February and August-December. However, the positive correlation coefficients of the NSQM showed a greater change than those of the SSQM, while the negative correlation coefficients of the SSQM showed a greater change than those of the NSQM. An increase in SPEI has an obvious role in promoting the annual NDVI on the NSQM and suppressing the annual NDVI on the SSQM (Chen et al., 2017; Hou et al., 2017; Qin et al., 2017).
In summary, vegetation on the NSQM is most sensitive to the overall hydrothermal dynamics of SPEI-4 in June, and March-June is the critical period for vegetation growth. Vegetation on the SSQM is most sensitive to the hydrothermal dynamics of SPEI-3 in May, and March-May (spring) is the critical period for vegetation growth. Vegetation growth on the NSQM and SSQM is mainly affected by hydrothermal dynamics in the shorter and medium timescales (3-7 months) and is weakly affected by hydrothermal dynamics in the longer timescales (11-12 months).

3.3 Spatial differences in vegetation response to hydrothermal dynamics on the NSQM and SSQM

3.3.1 Spatial distribution of NDVI and SPEI correlations on the NSQM and SSQM

To further explore the spatial differences in the vegetation growth response to hydrothermal dynamics, the correlation between the annual NDVI and SPEI of spring (SPEI-3 in May) was used to analyze the response of vegetation growth to hydrothermal dynamics on the NSQM and SSQM. The correlation between the annual NDVI and SPEI of spring on both NSQM and SSQM was significant; therefore, the correlation between the spring SPEI and annual NDVI was used as a basis.
Figure 5 shows that the spatial distribution of the NDVI and SPEI correlation coefficients in the QMs ranges from -0.84 to 0.88, with 90.34% of the areas in the QMs having positive correlations and only 9.66% of the areas in the QMs having negative correlations. The areas with negative correlations are mainly located in the gentle NSQM and SSQM, which may be related to serious destruction of vegetation due to urban expansion and to climatic factors. In conclusion, the rising SPEI in spring in most areas of the QMs significantly contributes to the growth of vegetation throughout the year.
Figure 5 Spatial distribution of correlation coefficients and significance between NDVI and SPEI
The average correlation coefficient between NDVI and SPEI on the NSQM is 0.24, with 81.05% of the areas on the NSQM being positively correlated and 23.29% of the areas on the NSQM being significantly positively correlated. The average correlation coefficient between NDVI and SPEI on the SSQM is 0.34, with 92.80% of the areas on the SSQM being positively correlated and 29.30% of the areas on the SSQM being significantly positively correlated. This indicates that wetting has a greater effect on promoting vegetation growth on the SSQM compared with the NSQM. It is worth noting that NDVI and SPEI in the triangle of “Taibai-Liuba-Foping” on the SSQM are significantly negatively correlated, indicating that spring wetting in this region suppresses vegetation growth throughout the year; and that the topography of this region is complex with high altitude and less human activity. Climatic factors are the main reason for vegetation change.
It has been pointed out that a large amount of vegetation (e.g., Abies fargesii) died in the high elevation areas of the QMs (Deng et al., 2018b), which may be related to spring wetting in the QMs from 2000 to 2018.

3.3.2 Differences in the response of major vegetation types to hydrothermal dynamics on the NSQM and SSQM

This study summarizes and reclassifies the vegetation on the NSQM and SSQM based on the vegetation cover types of the QMs and the vegetation cover classification scheme of the International Geosphere-Biosphere Programme (Bai et al., 2014; Yang et al., 2018) (Table 2). The spatial distribution of NDVI and SPEI correlation coefficients was extracted by mask according to the vegetation cover types on the NSQM and SSQM. NDVI and SPEI correlation coefficients for different vegetation cover types were obtained (Table 3), which indicate that growth of the main vegetation types on the NSQM and SSQM responded positively to hydrothermal dynamics—with grasses possessing the most sensitive response, followed by shrubs, broad-leaved forest, needle-leaved forest, and alpine meadow. Trees have well-developed root systems, and the effect of hydrothermal dynamics on woodland is less than that on grasses and shrubs. Due to the incorporation of anthropogenic influences (e.g., irrigation) in farmland vegetation, the sensitivity of the response to hydrothermal dynamics is reduced. The effective cumulative temperature in an alpine meadow may be an important factor in determining vegetation growth, so its sensitivity to hydrothermal dynamics is poor. The correlation coefficients of bare land were more contingent, which was mainly rocky and sparsely vegetated. Vegetation has the weakest response to hydrothermal dynamics in urban areas, which is mainly due to the most serious anthropogenic influences.
Table 2 Land cover reclassification
Vegetation type Reclassification Vegetation type Reclassification
Montane cold temperate evergreen
needle-leaved forest 		Needle-leaved forest Alpine and sub-alpine meadow Alpine meadow
Montane cold temperate deciduous
needle-leaved forest 		Typical meadow Grasses
Montane temperate evergreen
needle-leaved forest 		Temperate sparse shrubby-grass slopes
Warm evergreen needle-leaved forest Croplands Farmland crops
Montane deciduous broad-leaved forest Broad-leaved forest Snow and ice Water bodies
Typical deciduous broad-leaved forest Water bodies
Warm-temperate evergreen broad-leaved deciduous mixed forest Wetlands Wetlands
Temperate deciduous broad-leaved forest Construction land Urban areas
Alpine-subalpine evergreen deciduous shrubs Shrubs Barren or sparsely vegetated Bare land
Subalpine deciduous broad-leaved shrubs
Warm deciduous broad-leaved shrubs
Temperate deciduous shrubs
Temperate deciduous broad-leaved shrubs
Table 3 Correlation coefficients between NDVI and SPEI for different vegetation types on the northern and southern slopes of the Qinling Mountains
Grasses Shrubs Broad-leaved forest Needle-leaved
forest 		Alpine meadow Farmland crops Bare land Water bodies Urban areas
NSQM 0.35 0.35 0.33 0.30 0.20 0.06 0.23 0.09 -0.07
SSQM 0.37 0.33 0.31 0.28 0.19 0.31 0.09 0.18 -0.02
The response of the main vegetation types on the NSQM and SSQM to hydrothermal dynamics differ from one vegetation type to another. Grassland response to hydrothermal dynamics on the SSQM is more sensitive than that on the NSQM, because the root system of grasses is not well developed and can only use the surface soil moisture. Moreover, the density of grasses on the SSQM is larger than that on the NSQM, so grasses are more sensitive to moisture requirements. Shrubs, broadleaved forest, needle-leaved forest, and alpine meadow are all more sensitive to hydrothermal dynamics on the NSQM. The correlation coefficients of NDVI and SPEI were greater because the shrubs and broad-leaved forests on the SSQM are mostly evergreen, and the deciduous vegetation on the NSQM. Moisture was the dominant climatic factor for vegetation change, hence having even greater correlation coefficient. In summary, vegetation types on the NSQM are more sensitive to hydrothermal dynamics than vegetation types on the SSQM.

3.3.3 Vertical differences in vegetation response to hydrothermal dynamics

(1) Distribution of vertical differences in vegetation response to hydrothermal dynamics
To investigate the differences in vegetation response to hydrothermal dynamics on the NSQM and SSQM in terms of changes in elevation, the elevations of the NSQM and SSQM are divided into 50-m intervals for reclassification, and the spatial distribution maps of NDVI and SPEI correlation coefficients are extracted using masks in ArcGIS. Areas with elevations below 500 m are mainly urban agglomerations, where human activity is the main factor influencing vegetation change, so this part focuses on elevations above 500 m. Figure 6 shows the change in NDVI and SPEI correlation coefficients with elevation. The correlation coefficients between NDVI and SPEI on both the NSQM and SSQM increase first and then decrease with increasing elevation.
Figure 6 Correlation coefficients between NDVI and SPEI with altitude on the northern and southern slopes of the Qinling Mountains
Specifically, correlation coefficients between NDVI and SPEI with elevation were divided into six parts: 1) Elevations of 500-800 m: The correlation coefficients of both the NSQM and SSQM increase rapidly with elevation, indicating that vegetation is more and more sensitive to hydrothermal dynamics; but the correlation coefficients are larger on the SSQM than on the NSQM. 2) Elevations of 800-1200 m: The correlation coefficients between NDVI and SPEI for both the NSQM and SSQM reach their highest value, indicating that the vegetation response to hydrothermal dynamics is most sensitive on both the NSQM and SSQM in this elevation range. In addition, the correlation coefficients of the NSQM and SSQM are close to each other. 3) Elevations of 1200-1900 m: The correlation coefficients of both the NSQM and SSQM decrease rapidly with increasing elevation, and the sensitivity of vegetation to hydrothermal dynamics decreases with increasing elevation. The correlation coefficients of the NSQM are larger than those of the SSQM, and the vegetation response is more sensitive. 4) The correlation coefficients between NDVI and SPEI for vegetation at elevations of 1900-2500 m show a decreasing “slowing down” phase, remaining stable with changes in elevation; but the differences between the NSQM and SSQM are considerable, with the correlation coefficients of the NSQM increasing and those of the SSQM decreas-ing. 5) Elevations of 2500-3000 m: The correlation coefficients of both the NSQM and SSQM decrease rapidly with increasing elevation. 6) Elevations above 3000 m: The correlation coefficients between NDVI and SPEI on both the NSQM and SSQM fluctuate at low levels, indicating that the hydrothermal dynamics levels may not be the main limiting factor for vegetation growth at this elevation.
(2) Analysis of the causes of vertical differences in the vegetation response to hydrothermal dynamics in the QMs
Vegetation at elevations of 500-800 m in the QMs consists of mainly low crops, fruit trees, grasses, shrubs and evergreen broad-leaved forests. Compared to the SSQM, the NSQM is likely to experience faster expansion of urban agglomerations, resulting in weaker vegetation response to climate change; Compared to the NSQM, the SSQM has a higher vegetation cover, therefore, vegetation response to climate is stronger and the correlation coefficient is larger on the SSQM. The proportion of fruit trees, grasses, and shrubs increases with increasing elevation, and the proportion of low crops and other crops decreases, transitioning to deciduous broad-leaved forests (Zhu et al., 2009). The vegetation at elevations of 800-1200 m on the NSQM and SSQM comprises mainly shrubs, grasses, and deciduous broad-leaved forest (e.g., Quercus variabilis forest) (Zhu et al., 2009), which are considerably sensitive to moisture. Therefore, below 1200 m, the sensitivity of the vegetation NDVI to hydrothermal dynamics increases with increasing elevation. At elevations of 800-1200 m, vegetation growth is most sensitive to hydrothermal dynamics.
At elevations of 1200-1900 m, there is a gradual change from grasses, shrubs, and Quercus variabilis forest to natural forest, such as Sharptooth oak and Quercus liaotungensis forest, in the context of main vegetation type, so vegetation growth becomes less sensitive to hydrothermal dynamics. At elevations of 1800-2500 m, the correlation coefficients between NDVI and SPEI for vegetation show a slow “drop down” phenomenon, probably because, at this elevation range, a transitional zone forms gradually from deciduous broad-leaved forest to mixed coniferous forest, and the response of coniferous forest to hydrothermal dynamics is much weaker than that of broad-leaved forest. In addition, temperature brings about changes from inhibiting vegetation growth to promoting vegetation growth at elevations of 1600-2700 m (Chen et al., 2019). Such changes in the relationship between vegetation growth and temperature variation are more marked on the NSQM compared with the SSQM. This results in a joint beneficial effect from temperature and precipitation on vegetation growth. The correlation coefficients of NDVI and SPEI on both the NSQM and SSQM show a decrease with increasing elevation.
The transition from mixed coniferous and broad-leaved forest to coniferous forest (e.g., Abies fargesiiandLarix chinensis) occurs at elevations of 2500-3000 m. The water demand of coniferous forest is lower than that of broad-leaved forest, resulting in decreasing correlation coefficients between NDVI and SPEI with increasing elevation.
The main vegetation types above 3000 m are Larix chinensis and alpine meadow, probably because of the high altitude and, in turn, low temperature, as heat becomes the main limiting factor for vegetation growth (Chen et al., 2019). Therefore, the correlation coefficient between NDVI and SPEI was the lowest compared to other elevation bands.
In summary, the sensitivity of vegetation growth to hydrothermal dynamics in the QMs varies in the vertical direction. On the one hand, this may be due to variation in water and the heat distribution due to “mountain mass effect”. On the other hand, it may be due to variation in the vertical distribution of vegetation types.

4 Conclusions and discussion

4.1 Conclusions

(1) The overall vegetation cover of the QMs improved significantly from 2000 to 2018, with the increase in NDVI, as well as the area ratio, on the SSQM being higher than those on the NSQM, and vegetation improvement on the SSQM is better than that on the NSQM. The average annual change rate of NDVI on the NSQM was 0.017/10a, with 79.31% of the areas showing an increasing trend and 53.15% of the areas showing increasing significance. The average annual change rate of NDVI on the SSQM was 0.031/10a, with 93.92% of the areas showing an increasing trend and 77.39% of the areas showing significance.
(2) The overall wetting trend in the QMs was not significant, while the wetting rate and area ratio of the NSQM were greater than those of the SSQM. The annual SPEI had 96.63% of the areas in the QMs showing insignificant change. The average annual change rate of SPEI on the NSQM was 0.13/10a, and 87.18% of the areas an increasing trend in SPEI; the average annual change rate of SPEI on the SSQM was 0.022/10a, and 71.43% of the areas had an increasing trend in SPEI.
(3) Vegetation on the NSQM is most sensitive to the overall hydrothermal dynamics of SPEI-4 in June, and March-June is the critical period for vegetation growth; vegetation on the SSQM is most sensitive to the hydrothermal dynamics of SPEI-3 in May, and March-May (spring) is the critical period for vegetation growth. Vegetation growth on the NSQM and SSQM is mainly affected by hydrothermal dynamics on short and medium timescales and weakly responds to hydrothermal dynamics on long timescales, but vegetation on the NSQM is more susceptible to hydrothermal dynamics than that on the SSQM. Vegetation growth on the NSQM was mainly affected, severely, by the multi-scale hydrothermal dynamics in June, and vegetation growth on the SSQM was mainly affected, severely, by the multi-scale hydrothermal dynamics in May.
(4) Areas with a positive correlation between NDVI and SPEI accounted for 90.34% of the areas in the QMs. The sensitivity of vegetation to hydrothermal dynamics increases first and then decreases with increasing elevation. An elevation of 800-1200 m is the most sensitive range in terms of vegetation response. The sensitivity of the vegetation response decreases with increasing elevation from 1200 to 3000 m. Grasses are the most sensitive vegetation type to hydrothermal dynamics on both the NSQM and SSQM, but most vegetation types on the NSQM are more sensitive to hydrothermal dynamics than those on the SSQM.

4.2 Discussion

This paper systematically explores the sensitivity and spatial differences of vegetation response to hydrothermal dynamics on the NSQM and SSQM. The critical periods for the impact of hydrothermal dynamics on vegetation growth on the NSQM and SSQM are revealed. Analyses of the sensitivity of vegetation types to hydrothermal dynamics and the vertical differences in vegetation response are conducted. The vertical differences in vegetation response are also analyzed, which can reflect the pattern and mechanism characteristics of the main vegetation types with respect to their response to hydrothermal dynamics in warm temperate and subtropical regions. This can also help to elucidate the differences in vegetation response and vegetation vulnerability in the north-south transitional zone of China under climate fluctuations. The latter is important for improving the accuracy of terrestrial vegetation prediction in the context of climate change and for illuminating vegetation change patterns in different climate zones. Positive changes are occurring in the terrestrial ecosystems of the QMs in the context of global change. The results of this study will provide an important theoretical basis for vegetation conservation and management and the planning of forestry and agriculture in the QMs.
Some problems are found in this study, which we hope can be solved with the enrichment of research data in a follow-up study: (1) In terms of the spatial distribution, the percentage of areas where the effect of hydrothermal dynamics on vegetation reached significance was not very high. This may be due to the influence of data accuracy, time period, and spatial scope, as well as the fact that there are not many high-altitude meteorological stations in the Qinling area, meaning the available data are limited. Moreover, there may be a lag in the response of vegetation growth to hydrothermal dynamics due to the complex topography of the QMs, and the lag time may vary in spatial extent. (2) Some studies have shown that the influence of anthropogenic factors on the NDVI of vegetation has become increasingly pronounced (Deng et al., 2018a; Tang et al., 2019). Implementation of policies that return farmland to forest, for example, has led to a rapid increase in the NDVI in the Loess Plateau region (Zhao et al., 2017). This has quickly improved the region’s vegetation cover under the combined effect of policy and climate, and the areas where the policies were were implemented on returning farmland to forest land are mainly low and middle mountainous areas and low mountainous hilly areas (Zhang, 2017; Chen et al., 2019). Although the QMs are different from areas such as the Loess Plateau, the implementation of similar policies on returning farmland to forest land and grassland is needed there, in addition to climatic factors. However, while the implementation of such policies significantly improves regional vegetation cover, the feedback effects of the improved vegetation cover on the regional climate should be explored. For example, whether the impact of increased evapotranspiration on water resources will increase the probability of drought. Such feedback effects cannot be neglected. Moreover, how to quantify the impact of human activity on vegetation cover change should be focused on in follow-up studies.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
Bai H Y, Liu K, Wang J et al., 2019. Vegetation Response and Adaptation in the Qinling Mountains under Climate Change Background. Beijing: Science Press,16-18. (in Chinese)

[2]
Bai H Y, Ma X P, Gao X et al., 2012. Variations in January temperature and 0℃ isothermal curve in Qinling Mountains based on DEM. Acta Geographica Sinica, 67(11):1443-1450. (in Chinese)

[3]
Byakatonda J, Parida B P, Moalafhi D B et al., 2018. Analysis of long term drought severity characteristics and trends across semiarid Botswana using two drought indices. Atmospheric Research, 213:492-508.

DOI

[4]
Chen C N, Zhu L Q, Tian L et al., 2019. Spatial-temporal changes in vegetation characteristics and climate in the Qinling-Daba Mountains. Acta Ecologica Sinica, 39(9):3257-3266. (in Chinese)

[5]
Chen L, Li S H, Hou L et al., 2017. Response of Larix chinensis radial growth to climatic factors based on the Vaganov-Shashkin model. Chinese Journal of Applied Ecology, 28(8):2470-2480. (in Chinese)

[6]
Dai A G, 2012. Increasing drought under global warming in observations and models. Nature Climate Change, 3:52.

DOI

[7]
Deng C H, Bai H Y, Gao S et al., 2018a. Comprehensive effect of climatic factors on plant phenology in Qinling Mountains region during 1964-2015. Acta Geographica Sinica, 73(5):917-931. (in Chinese)

[8]
Deng C H, Bai H Y, Gao S et al., 2018b. Spatial-temporal variation of the vegetation coverage in Qinling Mountains and its dual response to climate change and human activities. Journal of Natural Resources, 33(3):425-438. (in Chinese)

[9]
Dong J W, Liu J Y, Zhang G L et al., 2013. Climate change affecting temperature and aridity zones: A case study in Eastern Inner Mongolia, China from 1960-2008. Theoretical and Applied Climatology, 113(3):561-572.

DOI

[10]
Fu B J, Yu D D, Lü N, 2017. An indicator system for biodiversity and ecosystem services evaluation in China. Acta Ecologica Sinica, 37(2):341-348. (in Chinese)

[11]
Gao J B, Jiao K W, Wu S H, 2019. Revealing the climatic impacts on spatial heterogeneity of NDVI in China during 1982-2013. Acta Geographica Sinica, 74(3):534-543. (in Chinese)

[12]
Hou L, Li S H, Chen L et al., 2017. Reconstruction of the historical temperature from February to April of the Qinling Mountains in recent 200 years. Geographical Research, 36(8):1428-1442. (in Chinese)

[13]
Hou M T, Zhao H Y, Wang Z et al., 2013. Vegetation responses to climate change by using the satellite-derived normalized difference vegetation index: A review. Climatic and Environmental Research, 18(3):353-364. (in Chinese)

[14]
Huang J P, Ji M X, Xie Y K et al., 2016. Global semi-arid climate change over last 60 years. Climate Dynamics, 46(3/4):1131-1150.

DOI

[15]
Hutchinson M F, Xu T, 2013. Anusplin Version 4.4 User Guide.

[16]
Kang M Y, Zhu Y, 2007. Discussion and analysis on the geo-ecological boundary in Qinling Range. Acta Ecologica Sinica, 27(7):2774-2784. (in Chinese)

[17]
Kong D D, Zhang Q, Gu X H et al., 2016. Vegetation responses to drought at different timescales in China. Acta Ecologica Sinica, 36(24):7908-7918. (in Chinese)

[18]
Li Y J, Ding J L, Zhang J Y et al., 2019. Response of vegetation cover to drought in the northern slope of the Tianshan Mountains during 2001-2015 based on the land-use and land-cover change. Acta Ecologica Sinica, 39(17):6206-6217. (in Chinese)

[19]
Liu S L, Tian Y Y, Yin Y J et al., 2016. Temporal dynamics of vegetation NDVI and its response to drought conditions in Yunnan province. Acta Ecologica Sinica, 36(15):4699-4707. (in Chinese)

[20]
Liu Z H, Tim R M, Van Niel T G et al., 2008. Introduction of the professional interpolation software for meteorology data: ANUSPLIN. Meteorological Monthly, 34(2):92-100. (in Chinese)

[21]
Lu F Z, Lu H Y, 2019. A high-resolution grid dataset of air temperature and precipitation for Qinling-Daba Mountains in central China and its implications for regional climate. Acta Geographica Sinica, 74(5):875-888. (in Chinese)

[22]
Mckee T B, Doesken N J, Kleist J, 1993. The relationship of drought frequency and duration of timescales. Eighth Conference on Applied Climatology, American Meteorological Society: Anaheim, California,174-184.

[23]
Meng Q, Bai H Y, Guo S Z, 2020. Spatial-temporal variation of precipitation in Qinling area in recent 50 years based on Anusplin. Research of Soil and Water Conservation, 27(2):206-212. (in Chinese)

[24]
Palmer W C, 1968. Keeping track of crop moisture conditions, nationwide: the new crop moisture index. Weatherwise, 21(4):156-161.

DOI

[25]
Qi G Z, Bai H Y, Meng Q et al., 2019. Climate change in the Qinling Mountains in spring during 1959-2018. Arid Zone Research, 36(5):1079-1091. (in Chinese)

[26]
Qin J, Bai H Y, Liu R J et al., 2017. Reconstruction of March-June mean air temperature along the timberline of Mount Taibai, Qinling Mountains, Northwest China, over the last 144 years. Acta Ecologica Sinica, 37(22):7585-7594. (in Chinese)

[27]
Stocker T F, Qin D H, Plattner M et al., 2013. Climate Change 2013:The Physical Science Basis. Cambridge: Cambridge University Press,866-871.

[28]
Tang J, Cao H Q, Chen J, 2019. Effects of ecological conservation projects and climate variations on vegetation changes in the source region of the Yangtze River. Acta Geographica Sinica, 74(1):76-86. (in Chinese)

[29]
Thuiller W, Lavorel S, Araújo M B et al., 2005. Climate change threats to plant diversity in Europe. Proceedings of the National Academy of Sciences of the United States of America, 102(23):8245-8250.

[30]
Tong S Q, Lai Q, Zhang J Q et al., 2018. Spatiotemporal drought variability on the Mongolian Plateau from 1980-2014 based on the SPEI-PM, intensity analysis and Hurst exponent. Science of The Total Environment, 615:1557-1565.

DOI

[31]
Vicenteserrano S M, Beguería S, Lópezmoreno J I, 2010a. A multiscalar drought index sensitive to global warming: The standardized precipitation evapotranspiration index. Journal of Climate, 23(7):1696-1718.

DOI

[32]
Vicenteserrano S M, Beguería S, Lópezmoreno J I et al., 2010b. A new global 0.5° gridded dataset (1901-2006) of a multiscalar drought index: Comparison with current drought index datasets based on the Palmer drought severity index. Journal of Hydrometeorology, 11(4):1033-1043.

DOI

[33]
Xu X, Xu Y, Sun Q Q et al., 2018. Comparison study on meteorological spatial interpolation approaches in Kangdian region of China. Journal of Central China Normal University (Natural Sciences), 52(1):122-129. (in Chinese)

[34]
Yang S Y, Meng D, Li X J et al., 2018. Multi-scale responses of vegetation changes relative to the SPEI meteorological drought index in North China in 2001-2014. Acta Ecologica Sinica, 38(3):1028-1039. (in Chinese)

[35]
Zhang Y, Bai H Y, Su K et al., 2018. Spatial variation of extreme temperature change on southern and northern slopes of Shaanxi section in Qinling Mountains during 1960-2013. Acta Geographica Sinica, 73(7):1296-1308. (in Chinese)

[36]
Zhang Y P, 2017. Vertical distribution and comparison of forests in the north and south slope of Qinling Mountains[D]. Xi’an: Xi’an University of Science and Technology. (in Chinese)

[37]
Zhao A Z, Zhang A B, Liu H X et al., 2017. Spatiotemporal variation of vegetation coverage before and after implementation of Grain for Green Project in the Loess Plateau. Journal of Natural Resources, 32(3):449-460. (in Chinese)

[38]
Zhu X Q, Liu K, Li J G et al., 2009. Analysis on vegetation-environment gradient correlation in Qinling Mountain based on GIS. Research of Soil and Water Conservation, 16(2):169-175. (in Chinese)

Options
Outlines

/

〈 〉