Effects of multiple environmental factors on WUE
An increasing EWUE trend was observed at the forest site with a change in climate (E1) which might be due to the increasing trend in precipitation (2.982 mm year−1) that enhanced carbon assimilation over the 30 years (Table 4). However, as precipitation decreased at the grassland and cropland sites (− 4.807 and − 2.338 mm year−1), it reduced carbon assimilation as well (Table 4), which in turn negatively affected EWUE. This relationship between precipitation and carbon assimilation is supported by the results of a previous study by Sun et al.30 and Zhao et al.32. The EWUE trend (E1) at the forest site in the first period (1981–2000, A1 in Fig. 1a) was lower than in the second period (2001–2010, A2 in Fig. 1a), indicating an increase in GPP due to rising temperature. The cropland and grassland sites (Fig. 1b, c) showed an increasingly negative trend in EWUE (in contrast to the forest site), as they were affected by climatic variables such as precipitation, temperature, and solar radiation (Table 4).
Due to rising CO2 (E2–E1), EWUE increased at all three sites in both study periods due to its structural effect (where plant growth increased due to higher CO2 concentration and changed the plant structure with increasing LAI)3,31 (Fig. 1). The EWUE trends decreased due to the effect of aerosol concentration (E3-E1) at grassland and cropland sites. ET and carbon assimilation decreased due to the higher concentration of aerosols, which reduced the solar radiation reaching the earth’s surface23,31 (Table S4). The effect of nitrogen deposition at the forest site decreased in the second period (2001–2010, A2 in Fig. 1a, Table S4) but continued to exhibit a negative trend.
Experiment E5 (ALL), which considered all climate and environmental factors, showed an increase in EWUE at the forest site, which was highly affected by CO2 fertilization and increased precipitation over the 3 decades (Fig. 1a, Table 4). However, at the grassland and cropland sites, EWUE decreased due to the negative precipitation trend and the positive trends in temperature and shortwave radiation over the 30-year period (Fig. 1b,c; Table 4). EWUE increased between the two study periods at the forest site but decreased at the grassland and cropland sites, likely due to the increase in CO2 and the effects of climatic variables.
In Fig. 2, the trends for TWUE (Case E1, ‘CLIM’) were steeper than those of EWUE, due to lower Tr than ET and higher GPP in the forest ecosystem (Figs. 2a, S3). However, the negative trend of TWUE was lessened due to the minimal Tr effect in ET at the grassland site, corroborated by ET and Tr trends (Figs. 2b, S4). The cropland site also showed a slight reduction in the negative trend of TWUE compared with EWUE due to higher positive trend in Tr compared with ET (Figs. 1c, 2c, S5). The CO2 fertilization effect (E2–E1) played an important role in the increase in carbon assimilation at the forest site and ultimately increased the TWUE trend compared with those at the grassland and cropland sites with statistical significance (p < 0.01) due to a higher LAI at the forest site2 (Fig. 2).
Aerosol concentration (E3–E1) and nitrogen deposition (E4–E1) had little effect on the variability of TWUE at all three sites compared with climate (E1) and CO2 concentration (E2-E1) (Fig. 2). The effect of aerosol concentration on TWUE was similar to that on EWUE at all three sites, as described previously. As for the effect of nitrogen deposition (E4-E1), the forest site showed a significant decrease (p < 0.01) in TWUE due to lack of nitrogen deposition for overall 3 decades (Fig. 2a, Table S4), but no significant trend was observed at the other two sites. The limiting effect of lack of nitrogen deposition on carbon assimilation corroborated the findings of previous studies15, 33.
Combining the effects of all variables on TWUE (E5), the trend increased at the forest site and decreased in the grassland and cropland, the same pattern as for the EWUE; however, the magnitude varied across sites due to the effect of Tr (Fig. 2). The first (1981–2000) and second (2001–2010) time periods exhibited the same trends for TWUE as for EWUE (increasing at forest and decreasing at cropland) except at the grassland site. It caused by abnormal behavior of aerosol concentration at the grassland site [which showed a decreasing trend in EWUE (Fig. 1b, E3–E1) and an increasing trend in TWUE (Fig. 2b, E3–E1)] and reduces the negative trend in TWUE for all factors (E5) in the second period compared with the first (Fig. 2b).
Figure 3 shows the IWUE at the three study sites. The effects of climate ‘CLIM’ showed opposite trends with respect to EWUE and TWUE. VPD, which is a function of air temperature, has a strong relationship with WUE. Here, the effects of VPD were removed from the analysis by introducing it only in IWUE15,26. This phenomenon explains the climate warming pattern especially at the forest and cropland sites, which directly affected plant growth by increasing VPD. The effect of CO2 fertilization (E2–E1) positively affected the IWUE due to an increase in atmospheric CO2 concentration.
In IWUE, the effects of aerosol and nitrogen deposition followed similar patterns to those of EWUE and TWUE (Fig. 3). The overall trend (E5) showed a decrease in IWUE at the forest site and increases in grassland and cropland sites due to the effects of climatic variables (Table 4). For both study periods, the trends in IWUE (E1 and E5) remained same at all three study sites but changed signs from positive to negative and vice versa compared with EWUE and TWUE due to removal of the VPD effect (Fig. 3).
Impact of climatic variables on WUE and its implications
Table 5 shows the partial correlations between multiple WUE terms and climatic variables (precipitation, solar radiation, and temperature). For the forest site (CN-Qia), EWUE and TWUE were positively correlated with all three climatic variables. Especially, the shortwave solar radiation showed a significant partial correlation with EWUE and TWUE (0.532 and 0.579, respectively, p < 0.05; Table 5). Solar radiation increased photosynthesis and carbon assimilation, but a higher LAI reduced the amount of radiation reaching the surface and ultimately reduced soil evaporation at the forest site15. Increased climate warming created drought-like conditions with decreasing precipitation trends (– 3.679 mm year−1) over the last 10 years of the study period at forest site (2001–2010, Table 4). However, deeper roots of the forest trees could withdraw water from greater depths and were less affected by the lack of precipitation10; this resulted in a positive EWUE trend (Fig. 1).
At the grassland site, EWUE was somewhat correlated with precipitation (0.4010, p < 0.05; Table 5) but not with solar radiation and showed a significant negative correlation with temperature (– 0.499, p < 0.05; Table 5). Precipitation at the grassland site showed a decreasing trend (– 4.807 mm year−1 for 1981–2010, Table 4) with relatively low mean annual temperature (~ 4.9 °C) and precipitation (400 mm, Table 1). These conditions represented a water stress environment with low LAI (due to less precipitation), which also reduced carbon assimilation. However, the last 10 years showed a little increase in the precipitation trend for grassland (Table 4), which did not affect the decreasing trend in carbon assimilation and ET and further decreased the negative EWUE trend possibly due to dry conditions (Fig. 1b).
For the cropland site, a fairly significant negative partial correlation was observed for EWUE with solar radiation (– 0.451, p < 0.1; Table 5). An increase in solar radiation caused higher ET (due to no water stress) and lower EWUE. This phenomenon was stronger for TWUE, which had a much higher negative partial correlation (– 0.826, p < 0.05; Table 5). A negative partial correlation was also observed with temperature, as rising temperature (0.008 K) increased ET at the cropland site which decreased EWUE and TWUE trends (Table 5, Figs. 1c, 2c). The decreasing trend of precipitation (– 2.338 mm year−1; Table 4) had little effect on EWUE and showed a non-significant partial correlation value (0.111), likely due to no water stress at the cropland site (Table 5).
For all three sites, IWUE showed opposite trends (positive to negative in case of forest site and negative to positive in case of grassland and cropland sites) to EWUE and TWUE because the effect of VPD (which is a function of temperature) was removed (Fig. 3). The results reflect a warming pattern at forest and cropland sites, and the grassland site showed an increasingly drier climate. Other climatic variables, such as specific humidity, longwave solar radiations, wind speed, and surface pressure, did not show significant trends at any site (Table 4).
Implications of WUE over multiple LCTs
WUE is the ratio between the water used in plant metabolism (or the amount of carbon uptake) to water loss from plants34. It is considered to be an important index for the study of the increase in atmospheric CO2 concentration, its effects on the ecosystem, and the changing climate with significant warming conditions35. Due to the increase in irrigation activities, the land use land cover changes have disturbed the carbon and water cycle. These disturbances to the terrestrial ecosystem increased the EWUE at the forest ecosystem and exhibited the greening of the land surface. Another reason might be an increase in atmospheric CO2 available to the plants which increases the rate of GPP2 (Fig. S3). However, in grassland and cropland conditions, the EWUE decreases that might have serious concerns for the warming conditions (Figs. S4, S5). In the forest ecosystem, the rate of water gain (carbon uptake) was higher than the rate of water loss by plants. However, the opposite trends were observed at the grassland and cropland sites (Figs. S4, S5). When the rate of water loss from plants was higher, the ecosystem moved to drier conditions, causing droughts in the long run. The increasing CO2 level also caused warmer conditions, which increased ET and led to water stress for plants.
The three LCTs in this study showed different trends in WUE under similar increasing trends for temperature and shortwave solar radiation over the last 3 decades. The different climate zones and soil characteristics associated with each site (Cfa-warm temperate for forest site, Bsk-arid for grassland, and Dfa-boreal for cropland, Table 1) affect their WUE trends. The forest and cropland sites showed a shifting of climate to warmer conditions (Table 4, Figs. S3 and S5); however, the grassland site showed a drier climatic pattern with significant decreasing precipitation trend and the increasing trend for temperature and solar radiations. Considering the effect of albedo, forests have a lower albedo than grassland and cropland ecosystems, which induced a summer cooling effect10,36. However, this (cooling effect) occurred due to the carbon loss to the atmosphere and reduction in carbon sink to terrestrial ecosystems from the atmosphere, which is reflected in increasing EWUE (E5) and TWUE (E5) trends at the forest site and decreasing trends at cropland and grassland sites37 (Figs. 1, 2). The anthropogenic increase of CO2 in the atmosphere, more than any other factor caused an increase in EWUE and TWUE (Figs. 1, 2), moreover, the reduction in carbon sink due to land cover changes leaves more carbon in the atmosphere and induced further warming38. The increased temperature due to water stress can lead to a reduction in soil moisture and ET, which can increase vegetation mortality13,39,40.
Forest and cropland sites respond differently to water stress. In drought conditions, forests can have green canopies for longer periods than croplands because trees can access deeper stored soil water; this ultimately reduces the effects of drought and high temperature over forests10. Here, the two study periods (1981–2000 and 2001–2010) showed increasing significant trends for temperature and shortwave solar radiations at all three sites (Table 4). At the forest site, the second period (2001–2010) showed higher trends for ‘E5’ in EWUE and TWUE than the first period (1981–2000) due to elevated CO2 in the atmosphere, which increased carbon assimilation with the increase in temperature (Figs. 1a, 2a, S3; Table 4). The grassland site showed a decrease in EWUE (E5) in the second period (2001–2010) compared with the first (1981–2000) due to negative trends of climate effect and aerosol concentration (Figs. 1b, 2b) as explained in the previous section. The cropland site (Figs. 1c, 2c) showed a decrease in EWUE (E5) and TWUE (E5) trends in the second period (2001–2010) compared with the first period (1981–2000) because of lower carbon assimilation and higher ET caused by the rise in temperature (0.027 K in the first period and 0.173 K in the second period; Table 4). EWUE and TWUE at all sites (Figs. 1, 2) showed their maximum trends (increasing/decreasing for forest/grassland and cropland, respectively) for E5 during the 2001–2010 period due to increased warming induced by rising CO2 in the atmosphere and other climatic variables38.
Due to increasing temperature, VPD and demand for ET both increased29. This phenomenon is also supported by Fig. 3, in which the IWUE exhibits opposite trends (compared with EWUE and TWUE in E1 and E5 experiments; positive to negative trends in case of forest site and negative to positive trends in case of grassland and cropland sites) for all sites when the effect of VPD was removed. These results have implications for climate due to rising CO2 and different feedback mechanisms to the atmosphere from various LCTs.
Potential uncertainties and limitations
Quantification of uncertainty in earth system models for simulation of ecological processes plays a critical role in the authentication of results41. The sources of uncertainty in LSMs include model uncertainty, climate data (forcing) uncertainty, and initial conditions uncertainty42,43. The LSM used in this study was the CLM5.0, which was updated regularly on the basis of model parameterization and uncertainties previously defined by scientists; however, multiple processes involved in the carbon and water cycles can generate additional uncertainties44,46,45,47. Figure S2c shows how a model estimation of GPP over cropland could be affected by an early onset of the spring season (model structural uncertainty). The grid-scale forcing data GSWP3v1 were used in this study, which also introduced uncertainties in the estimation of carbon and water cycle fluxes. This was corroborated by previous literature48,49. Figure S2b shows that the GPP estimation from the model was sensitive to precipitation (forcing data uncertainty). The carbon cycle requires a large number of spin-up cycles to reach equilibrium, and this affected the initial values of the parameters, which added to the overall uncertainty of the model43. In land carbon uptake, model structural uncertainty is predominant among all possible uncertainties41. Lovenduski and Bonan50 explained that there is a limit below which the uncertainty cannot be further reduced.
Limitations are common in modeling studies. Initially, the coarse resolution of forcing data (climatic variables) from GSWP3v1 was 0.5°. High resolution data or weather station/flux tower data are more reliable for site level studies. Due to non-availability of long-term site level data (1981–2010), gridded 0.5° GSWP3v1 data were used in the study. Secondly, very few studies have used the bio-geochemical cycle of CLM5.0, which needs to be analyzed further to enhance the accuracy of the model, especially at cropland regions. The site scale of the study was also one of the limitations, as the regional or global scale could better analyze the effects of multiple land cover types on the climate due to rising CO2 and the influence of nitrogen deposition and aerosol concentration. Future studies should be conducted at a regional to global scale to study this phenomenon in detail over multiple regions.
Source: Ecology - nature.com
