-
Over the TP, WUE decreases from southeast to northwest (Fig. 1). Large differences exist among the individual models regarding the spatial distribution of WUE, and the annual mean value of simulated WUE ranges from 0.235 g C (kg H2O)–1 (VISIT) to 1.706 g C (kg H2O)–1 (LPJ-wsl). In addition, the result of the ENSEMBLE is 0.923±0.522 g C (kg H2O)–1 (Table 1), indicating that the WUE for the TP simulated from the single models has great uncertainty. The estimated annual average WUE based on satellite remote sensing data is 0.465 g C (kg H2O)–1, and the MTE and FLUXCOM results are 0.897 g C (kg H2O)–1 and 0.657 g C (kg H2O)–1, respectively. This is mainly due to the uncertainty in the ET and GPP estimates on the TP between different data and models, which increases the uncertainty in WUE. The multi-model ensemble strategy can effectively reduce the uncertainty of the mode simulation WUE.
Figure 1. Annual mean water use efficiency (WUE) from the MODIS, MTE, FLUXCOM, and MsTMIP models (SG3 simulation) over the TP for 2001–10. ENSEMBLE is the ensemble mean of the nine MsTMIP models.
Model WUE g C (kg H2O)–1 CLM4 1.283 CLM4VIC 1.118 DLEM 0.816 ISAM 0.608 LPJ-wsl 1.706 ORCHIDEE-LSCE 1.552 SiB3 0.339 SiBCASA 0.650 VISIT 0.235 ENSEMBLE 0.923±0.522 MODIS 0.465 MTE 0.897 FLUXCOM 0.657 Table 1. Annual mean values from MsTMIP (SG3), MODIS, MTE, and FLUXCOM for the Tibetan Plateau (TP) from 2001 to 2010.
-
Figure 2 shows interannual time series of WUE, GPP, and ET from the MTE, FLUXCOM, MODIS, and the ensemble mean of the nine MsTMIP models. CO2 fertilization efficiency promotes the photosynthesis of vegetation, thereby increasing the value of GPP and improving the carbon-water coupling in the terrestrial ecosystems [the differences of annual GPP and WUE between SG3 and SG2 are 15.34 g C m–2 yr–1 and 0.09 g C (kg H2O)–1, respectively]. The annual WUE of SG2 is 0.19 g C (kg H2O)–1 less than that of SG1, which indicates that LULCC decreases the WUE. The trends in the different data indicate a significant upward trend in the WUE on the TP. We further analyzed the WUE trend values from different subregions and different plant functional types (PFTs), and found the largest trend values for the Brahmaputra [0.0049 g C (kg H2O)–1 yr–1] and the forest [0.0142 g C (kg H2O)–1 yr–1]. LULCC causes the trend value of WUE to decrease [Table 2, the trend of WUE of SG2 is 0.0008 g C (kg H2O)–1 yr–1 less than that of SG1], and the CO2 fertilization greatly improves the growth trend of WUE [Table 2, the trend of WUE of SG3 is 0.0023 g C (kg H2O)–1 yr–1 more than that of SG2].
Figure 2. Interannual changes of WUE, gross primary productivity (GPP), and evapotranspiration (ET) from the MTE (black line), FLUXCOM (green line), MODIS (orange line), and the ensemble mean of the nine MsTMIP models: SG1 (red), SG2 (blue), SG3 (purple). The gray lines represent the variables of the single models from SG3. The anomalies were calculated as the difference between annual value and the long-term mean over the period for each data source.
Dataset WUE [g C
(kg H2O)–1 yr–1;
from 1981 to 2010]WUE [g C
(kg H2O)–1 yr–1;
from 2001 to 2010)]SG1 0.0030** −0.0080** SG2 0.0022** −0.0071** SG3 0.0045** 0.0001 MODIS − 0.0121** MTE − 0.0050 FLUXCOM − 0.0008 Table 2. The linear trends of WUE from the MODIS (from 2001 to 2010), MTE (from 2001 to 2010), FLUXCOM (from 2001 to 2010), and MsTMIP models (SG1, SG2, and SG3; from 1981 to 2010) on the TP. ** indicates that the trends are statistically significant (p < 0.01).
-
Figure 3 shows the spatial distributions of the impacts of different climatic factors (temperature, precipitation, and radiation) on WUE. It is found that temperature dominates the change of WUE on the TP, and precipitation and radiation have little influence on WUE (Fig. 3d). This indicates that WUE is mainly restricted by thermal conditions on the TP. The annual average temperature is low on the TP, and the increase in temperature can alleviate the thermal stress of vegetation growth and increase the rate of photosynthesis (Chen et al., 2013; Xu et al., 2016). In the midwest and northwest of the TP, precipitation is the dominant factor and positively correlates with WUE. While in other areas, there is a negative correlation, similar to previous research (Guo et al., 2019; Fang et al., 2020).
Figure 3. The spatial distributions of the impacts of (a) temperature, (b) precipitation, (c) incoming shortwave radiation, and (d) the dominant climatic factors on WUE from the MsTMIP models on the TP for 1981–2010.
Figure 4 shows the spatial distribution of the WUE response rate to LULCC on the TP from 1981 to 2010. The ENSEMBLE indicates that LULCC makes a negative contribution to WUE in most areas of the TP but a significant positive contribution to the central and northwestern areas. For single models, the percentage of the simulated WUE response to LULCC varies greatly among the models. We believe that it is caused by the great uncertainty in the impact of the different LULCC models on GPP and ET over the TP. For the whole TP, LULCC resulted in a decrease in the annual average WUE, with a contribution rate of –20.63%, especially on the eastern plateau. We find that the negative impact of LULCC on WUE is mainly due to the negative contribution of LULCC to GPP.
Figure 4. The spatial patterns of the WUE response rate (%) to land use and land cover change (LULCC) on the TP from 1981 to 2010.
Figure 5 shows the response percentage of WUE to the CO2 fertilization effect on the TP. All models indicate a positive effect of CO2 fertilization on WUE in most areas of the TP. The positive effect is caused by the positive contribution of the CO2 concentration change to GPP and the negative contribution to ET. For the whole TP, the annual average WUE showed a positive response to CO2 fertilizatio- n (11.65%), but it was negative in the west and the northwest. It is worth noting that the positive contribution of the CO2 fertilization effect can offset part of the decrease in WUE caused by LULCC, but both contributions are much smaller than that of climate change.
-
The ecosystem resilience represents the ability of an ecosystem to sustain or recover its structure and functions with disturbances from hydroclimate. A resilient ecosystem could maintain or increase WUE to ensure its productivity in a water-limited environment due to drought. We further analyze the impact of human activities on the ecological resilience, which helps us understand how human activities affect the degree of carbon-water coupling under drought conditions.
Our research shows that the increase in atmospheric CO2 concentration greatly increased the ecological resilience over the TP, while LULCC reduced the ecological resilience over the central TP (Fig. 6). At the same time, studies have shown that as the climate warms and atmospheric CO2 rises, the Qinghai-Tibet Plateau is becoming greener. Conversely, grasslands are more vulnerable to drought than other PFT. Our research shows that the effect of CO2 fertilization can improve the ecological resilience of the TP meadow.
Figure 6. The spatial patterns of ecosystem resilience (Rd) to drought from the ensemble mean of the nine MsTMIP models (SG1, SG2, and SG3) for 1981–2010. Resilient: Rd ≥ 1; slightly non-resilient: 0.9 < Rd < 1; moderately non-resilient: 0.8 ≤ Rd ≤ 0.9; and severely non-resilient: Rd < 0.8.
We further analyzed the impact of LULCC and CO2 fertilization effects on WUE under drought years and average annual conditions. Under drought, the ecosystem WUE is clearly more sensitive to LULCC and CO2 fertilization effects (Fig. 7).
Model | WUE g C (kg H2O)–1 |
CLM4 | 1.283 |
CLM4VIC | 1.118 |
DLEM | 0.816 |
ISAM | 0.608 |
LPJ-wsl | 1.706 |
ORCHIDEE-LSCE | 1.552 |
SiB3 | 0.339 |
SiBCASA | 0.650 |
VISIT | 0.235 |
ENSEMBLE | 0.923±0.522 |
MODIS | 0.465 |
MTE | 0.897 |
FLUXCOM | 0.657 |