Variability of spring ecosystem water use efficiency in Northeast Asia and its linkage to the Polar-Eurasia pattern

Given that water use efficiency (WUE) is an important indicator to measure the trade-off between carbon uptake and water consumption within the ecosystem, better understanding the variation of ecosystem WUE and related driving factors is of great interest. In this study, the variability of spring ecosystem WUE in Northeast Asia (NEA) was investigated. The results show that its primary mode exhibits a monosign variation. This mode is directly controlled by the variability of gross primary productivity. The climate conditions also play remarkable roles, featuring that warm surface air temperature (high soil moisture) favors enhanced ecosystem WUE in northern (southern) NEA. Further analysis reveals that the Polar-Eurasia (POL) pattern can significantly impact the variability of spring ecosystem WUE in NEA through changing surface air temperature and soil moisture. When the POL pattern lies in the positive phase during spring, anticyclonic circulation anomalies with an equivalent barotropic structure prevail over northern NEA, concurrent with anomalous easterlies over southern NEA and a weakening of the East Asian jet (EAJ). Accordingly, anomalous downward motion is introduced over northern NEA, resulting in higher surface air temperature which is beneficial for the increase of local ecosystem WUE. Meanwhile, the easterly anomalies help to increase water vapor transport into southern NEA and the weakened EAJ can induce anomalous ascending over southern NEA, favoring the increase of precipitation and hence soil moisture, which consequently enhances the ecosystem WUE in southern NEA.


Introduction
Terrestrial ecosystem plays a vital role in modulating global water and carbon cycle through various physical and biological processes . For example, the terrestrial ecosystem affects water cycle by consuming water through evapotranspiration (ET) (Tang et al 2014), and influences carbon cycle by assimilating atmospheric CO 2 via photosynthesis (Liu et al 2020b). The trade-off between carbon uptake and water consumption can be expressed as ecosystem water use efficiency (WUE), which is usually defined as the ratio of gross primary productivity (GPP) to ET (Beer et al 2009). The WUE is an important ecosystem property for monitoring the changes in carbon and water exchanges (Wang et al 2018). Comprehensive understanding of ecosystem WUE variation and its related driving factors helps to assess the vulnerability and adaptability of ecosystem under changing climates (Huang et al 2021, Zhao et al 2022. Surface air temperature, solar radiation, soil moisture, vapor pressure deficit (VPD), precipitation, and terrestrial water storage are potential factors to drive WUE changes (e.g. Liu et al 2015, Zhang et al 2017. Nevertheless, the relationships between the ecosystem WUE and those factors are complex and vary by regions. In general, temperature is more strongly associated with the spatial gradient of annual WUE north of 50 • N, while precipitation shows greater contribution in the temperate and tropical regions (Sun et al 2016, Huang et al 2021. The climate controls of ecosystem WUE also differ for different vegetation types. For instance, precipitation is the dominant factor for the WUE of woodlands, croplands and grasslands in China's Loess Plateau, whereas temperature primarily influences the WUE of shrublands (Zhang et al 2016). The lagged effect of soil moisture on the WUE of forest remains longer for that of sparse vegetation and shrublands (Ji et al 2021). In addition, the impact of meteorological factors on the ecosystem WUE changes with seasons (Huang et al 2016). Compared to summer, rising temperature is more conducive to promoting the European forest WUE in early spring and late autumn (Montibeller et al 2022).
Northeast Asia (NEA) is one of the world's major economic centers, with dense population and diverse ecological patterns , Mo et al 2019. It is also a hot spot for the research of climate change and ecogeographic process, due to evident land surface warming (Dong et al 2016, Liu et al 2018. The ecosystem WUE in this region, on one hand, is closely related to the local grain yield (He et al 2010). On the other hand, it is an important indicator to assess the land degradation and rehabilitation (Kang et al 2020). Thus, it is of significance to investigate the variation of ecosystem WUE in NEA. Due to the warming, the NEA area has experienced a greening in spring since the 1980s (Piao et al 2015, Lian et al 2020. Although several studies have investigated the spatiotemporal variation of annual ecosystem WUE in NEA and its responses to meteorological factors (Sun et al 2016, Wang et al 2018, the variability of spring ecosystem WUE in this region and its controlling mechanisms remain open, which is the motivation of the present study.

Data and methods
The observation-based GPP and latent heat (LH) dataset with a spatial resolution of 0.5 • × 0.5 • for the period 1982-2011, which was upscaled from the eddy-covariance measurements at the flux tower sites using a model tree ensemble machine learning technique (Jung et al 2011), is supplied by the Max Planck Institute for Biogeochemistry (MPI-BGC). This dataset owns relatively small uncertainty and has been widely used (Zhang and Ye 2021). To complement the analysis and for comparison, the ensemble mean of the GPP and ET of the CLM5.0 (Lawrence et al 2019), IBIS (Yuan et al 2014), ISAM (Meiyappan et al 2015), and OCN (Zaehle and Friend 2010) simulations from the Trends and Drivers of the Regionalscale Sources and Sinks of Carbon Dioxide Project (TRENDY) version 10 for the period 1982-2020 are also used, in which the four dynamic global vegetation models were forced by the same forcing including the observed atmospheric CO 2 concentration, climate, and land use on global water and carbon cycles (Friedlingstein et al 2022).
Other datasets used include: (1) the ERA5 reanalysis data from the European Centre for Medium-Range Weather Forecasts, including 2 m temperature, 2 m dewpoint temperature, total cloud cover, zonal and meridional winds, vertical velocity, geopotential height, specific humidity, surface heat flux, soil moisture, and precipitation, with a horizontal resolution of 0. The ecosystem WUE is defined as the ratio between GPP and ET, in which the ET was calculated by multiplying the LH by a factor of 0.035 (Beer et al 2009, Tang et al 2014. The VPD at 2 m above the land surface, reflecting the atmospheric dryness stress (Liu et al 2020a), was calculated following Du et al (2018): where T 2m and T d are surface air temperature and dewpoint temperature at 2 m, respectively. The vertically integrated water vapor flux (WVF) was calculated in terms of WVF = − ∫ q V dp/g, in which q is specific humidity, V is horizontal wind vector, p is pressure, and g is gravity acceleration. Given that the water vapor content above 300 hPa is very little and the humidity measured at those levels is subject to large instrumental errors, the WVF was integrated from the surface to 300 hPa. Its divergence was derived by ∇ · WVF. The empirical orthogonal function (EOF) analysis was performed to identify the primary mode of spring (March-May) ecosystem WUE in NEA (35 • -55 • N, 105 • -140 • E). The wave activity flux (WAF), based on the equation proposed by Takaya and Nakamura (2001), was applied to investigate the Rossby wave energy propagation. Correlation and regression analyses, with the statistical significance determined by the student's t test, were also used. All the time series were detrended prior to the analysis. WUE in NEA. Given that the climatological ecosystem WUE in the sparsely vegetated regions shows very low values (less than 0.4 g C kg −1 H 2 O) during spring (figure S1), the bare and sparse vegetation areas were not considered in the analysis. As shown in figure 1(a), the EOF1 mode, which explains 20.63% of the total variance, exhibits a monosign pattern in NEA. This pattern indicates that the variation of spring ecosystem WUE is homogenous across the study area. The ecosystem WUE variability directly depends on the variations of GPP and ET (Huang et al 2015). Due to the coupled photosynthesis and transpiration, the GPP and ET may increase simultaneously under certain ambient conditions (Montibeller et al 2022). Greater increase in GPP than that in ET leads to enhanced ecosystem WUE, and vice versa (Liu et al 2020b). Yang et al (2016) revealed that the variability of annual WUE in arid regions is mostly controlled by the ET, whereas that for semi-arid/subhumid ecosystems is primarily regulated by the GPP.

Variability of spring ecosystem WUE in NEA
To determine relative contribution of the GPP and ET in controlling the variability of spring ecosystem WUE in NEA, we have calculated the regressions of GPP and ET with the PC1 series, which are shown in figures 1(c) and (d), respectively. It appears that positive anomalies in both GPP and ET dominate NEA. In comparison, the positive anomalies of GPP are more significant than those of ET. Furthermore, the GPP shows a higher coupling strength to the area-averaged ecosystem WUE variation in NEA (figures 1(e) and (f)), which is in agreement with the previous studies for mid-high latitude ecosystems (Wang et al 2018, Huang et al 2021. These findings suggest a more important role of GPP in the variability of spring ecosystem WUE in NEA. Similar results can be obtained if we use the TRENDY data and extend the time period to 2020 (figure S2).
It is noted that the PC1 series shown in figure 1(b) presents both interannual and interdecadal variations, with the decadal signal more prominent than the interannual signal. We further filtered the decadal variability of spring ecosystem WUE and re-performed the EOF analysis. After the decadal variability is filtered, the EOF1 mode of spring ecosystem WUE in NEA still displays a monosign pattern and this pattern is dominantly controlled by the interannual variability of GPP (figure S3), which are similar to the results from figure 1.

Associated local climate conditions
The above analyses illustrate that the variation of spring ecosystem WUE in NEA is primarily controlled by the GPP anomalies. Climate factors can  (2015) indicated that the increase of temperature and solar radiation may enhance the GPP and hence increase the ecosystem WUE at high latitudes. Yang et al (2016) found that the drought-induced water deficit may result in a decrease in GPP and then reduces the ecosystem WUE in semi-arid/sub-humid regions. In the following, we attempt to figure out the contribution of local climate factors to the variability of spring ecosystem WUE in NEA. Figure 2 shows the anomalies of surface air temperature, solar radiation, VPD, precipitation, and soil moisture regressed against the PC1 series in spring. Interestingly, a general out-of-phase variation is observed between the northern and southern parts of NEA. In northern NEA, the increased ecosystem WUE corresponds to the positive anomalies of surface air temperature, solar radiation, and VPD, while negative anomalies in precipitation and soil moisture. It suggests that the variation of spring ecosystem WUE in this region is largely driven by energy supply rather than water availability. The low dependence of ecosystem WUE on the water availability may be explained in two ways. On one hand, the northern NEA region is covered by the forests with extensive root system (figure S1(b)), which generally possess a greater tolerance to water stress by absorbing moisture from deeper soil layer (Huang et al  2021). On the other hand, snowmelt in northern NEA during spring may provide sufficient soil water . It is noteworthy that abnormally high soil moisture can decrease local surface temperature through modulating diabatic heating to the atmosphere (Koster et al 2016), which in turn inhibits the ecosystem WUE.
Regionally averaged over northern NEA (45 • -55 • N, 120 • -140 • E), the surface air temperature is most significantly correlated to the GPP and ecosystem WUE among the above factors; its correlations with the MPI-BGC (TRENDY) GPP and ecosystem WUE are 0.71 (0.82) and 0.59 (0.80), respectively, exceeding the 0.01 significance level. Thus, the consequence of high surface air temperature over northern NEA tends to increase the local GPP and hence the ecosystem WUE.
By contrast, in southern NEA, the positive PC1 values are associated with the negative anomalies of surface air temperature, solar radiation, and VPD, whereas positive anomalies in precipitation and soil moisture (figure 2), which hints that the variation of spring ecosystem WUE in this area is mainly driven by water availability rather than energy supply. One possible reason is that this region is mainly covered by temperate cropland and grassland (figure S1(b)), which cannot access deep water reserves to relieve water stress (Bastos et al 2020). The correlations between the soil moisture and the MPI-BGC GPP and ecosystem WUE averaged over southern NEA (35 • -40 • N, 108 • -120 • E) are 0.51 (p < 0.01) and 0.35 (p < 0.1), and the counterparts with the TRENDY GPP and ecosystem WUE are 0.43 (p < 0.01) and 0.34 (p < 0.05), respectively. That is to say, high soil moisture is beneficial for the increase in GPP and ecosystem WUE in southern NEA. It should be noted that the GPP is less sensitive to the VPD (the MPI-BGC and TRENDY GPP correlated with the VPD at −0.37 and −0.26, respectively) than to the soil moisture, supporting the findings by Liu et al (2020a) that soil moisture controls the dryness stress on ecosystem production.
In brief, the dominant climate factors affecting the variability of spring ecosystem in northern and southern NEA are different. In northern NEA, the enhancement of spring ecosystem WUE is more closely linked to the increase in surface air temperature, whereas that in southern NEA is more closely related to the increase of soil moisture.

Role of the POL pattern
It is noteworthy that large-scale atmospheric teleconnection patterns can affect terrestrial ecosystem through substantial influences on regional climate factors (Zhou et al 2013, Li et al 2016. The POL pattern, known as an atmospheric teleconnection mode characterized by seesaw variations in geopotential height over the polar region and the NEA region (Barnston and Livezey 1987), can give rise to abnormal rainfall and air temperature (Piao et al 2018, Gao et al 2019, Li et al 2020. An emerging question is whether the POL pattern plays a role in the variability of spring ecosystem WUE in NEA. To answer this question, figure 3(a) shows the normalized time series of the spring POL pattern index and the PC1 of the ecosystem WUE in NEA derived from the MPI-BGC and TRENDY datasets. It is of interest to find that the variation of the POL index is generally consistent with that of the PC1 series. The correlations between the POL index and the MPI-BGC and TRENDY PC1 series reach 0.57 and 0.56 (p < 0.01), respectively. Such an in-phase relationship can also be proved by the regression maps of the GPP and ecosystem WUE with the POL index, which display that the positive phase of spring POL pattern is associated with an increase of both GPP (figures 3(b) and (c)) and ecosystem WUE (figures 3(d) and (e)) in NEA. Moreover, the spatial distribution of the regression pattern between the ecosystem WUE and the POL index bears high resemblance to the EOF1 mode of spring ecosystem WUE in NEA (figures 1(a) and S2(a)). Therefore, the POL pattern is hypothesized to be an important large-scale atmospheric signal influencing the variability of spring ecosystem WUE in NEA.
How does the POL pattern exert influences on the ecosystem WUE in NEA during spring? To address this issue, we have checked the atmospheric circulation anomalies in association with the spring POL pattern over the course of 1982-2020. As shown in figure 4(a), the POL-related horizontal wave activity propagates from the polar region northeasterward to the northern NEA region. Accordingly, anticyclonic circulation anomalies (positive height anomaly) are introduced over northern NEA a response to the positive POL phase, which appear from the upper troposphere to the lower troposphere (figures 4(b)-(d)). These characteristics are noticed to resemble those associated with the PC1 series of spring ecosystem WUE in NEA (figures S4 and S5). Under the background of such atmospheric circulation anomalies, anomalous descending is formed in northern NEA ( figure 5(c)). The descending anomaly, on one hand, is conducive to adiabatic heating (Hu et al 2021, Sun et al 2021), and on the other hand helps to inhibit cloud generation and thus decrease the total cloud cover (figure 6(a)). The decreased cloud cover favors more incoming solar radiation (figure 6(b)) and causes positive anomalies in surface net heat flux (figure 6(c)). All of these provide favorable conditions for higher surface air temperature in northern NEA (figure 6(d)), consequently leading to an enhancement of local GPP and hence the ecosystem WUE. In fact, the positive phase of spring POL pattern is conducive to more incidences of atmospheric blocking events over this region (figure S6).
Due to the anomalous anticyclonic circulation residing over northern NEA, the southern NEA region is in the control of easterly anomalies in the upper and lower troposphere (figures 4(b) and (d)), favoring the water vapor transport from the Pacific Ocean into southern NEA. As a result, the water vapor converges in southern NEA ( figure 5(b)), providing abundant water vapor as required for the occurrence of precipitation. In addition, the anticyclonic circulation anomaly over northern NEA is coupled with an anomalous cyclonic circulation to its southwest side in the upper troposphere ( figure 4(b)), forming alternative positive-negative-positive anomalies of zonal wind from high latitudes to lower latitudes ( figure 5(a)). The pattern with the easterly anomalies around 30 • -50 • N and the westerly anomalies to its south suggests a weakening and southward displacement of the East Asian jet (EAJ), since the active region of the EAJ is climatologically located around 30 • -35 • N. As is known, the EAJ plays a dynamical role in the vertical motion by evoking a secondary circulation (Ding 2005). Due to the weakening of the EAJ, anomalous ascending is evoked on the left-hand side of the jet axis, where the southern NEA is right situated (figure 5(d)). The ascending anomaly provides a favorable dynamic condition for the occurrence of precipitation. Therefore, under the conditions of enhanced water vapor and anomalous ascending motion, the precipitation is inclined to increase in southern NEA (figure 6(e)), which further enhances the soil moisture in situ (figure 6(f)). As expected, the enhancement of soil moisture is conducive to the increase in local GPP and thus the ecosystem WUE in southern NEA.
To sum up, the POL pattern plays a pronounced role in the variability of ecosystem WUE in NEA during spring. The POL-related changes in the large-scale atmospheric circulations can increase the surface air temperature over northern NEA and the soil moisture over southern NEA, consequently resulting in the enhancement of GPP and ecosystem WUE.

Conclusions
Based on the MPI-BGC and TRENDY data, we have examined the variability of spring ecosystem WUE in NEA. It is indicated that the EOF1 mode of spring ecosystem WUE exhibits a monosign variation in NEA, and this pattern is mainly determined by the variability of GPP.
The dominant climate conditions controlling the variability of spring ecosystem WUE in NEA were further investigated. In particular, the role of the POL pattern was highlighted. The results show that the enhanced ecosystem WUE in northern NEA is more closely linked to the warming of surface air temperature, whereas that in southern NEA is more tightly associated with the increase of soil moisture. The POL pattern is expected to play a pronounced role via changing the surface air temperature and soil moisture. The positive phase of spring POL pattern can induce equivalent barotropic anticyclonic circulation anomalies over northern NEA, which are accompanied with anomalous easterlies over southern NEA and a weakening of the EAJ. As a response to the equivalent barotropic anticyclonic circulation anomalies, anomalous descending are dominant over northern NEA, resulting in more incoming solar radiation and surface net heat flux. These conditions favor higher temperature over northern NEA, thereby contributing to the increase of local GPP and ecosystem WUE. In southern NEA, the easterly anomalies can enhance the water vapor transport from the Pacific Ocean into the region, and anomalous ascending motion is arouse by the weakened EAJ, favoring the occurrence of precipitation and hence increasing the soil moisture. The increased soil moisture is conductive to enhancing the GPP and ecosystem WUE in southern NEA. The opposite situations are applicable for the negative phase of spring POL pattern.
In general, the findings obtained in this study are encouraging for a better understanding of the variability of ecosystem WUE in NEA. Certainly, the current study just presents a preliminary explanation and focuses on the impact of spring POL pattern. It is worth noting that the variation of the POL index is not consistent with that of PC1 series in some years, such as in 1997, 2001, and 2005 (see figure 3(a)). This reminds us that other factors, including other teleconnection patterns, may play a role in the variability of spring ecosystem WUE in NEA and hence offset the effect of the POL pattern. This issue deserves further investigation in the future studies.