Variation in Water Supply Leads to Different Responses of Tree Growth to Warming

: 14 Background: Global climate change, which includes changes in precipitation, 15 prolonged growing seasons, and drought stress caused by overall climate warming, is 16 putting increased pressure on forest ecosystems globally. Understanding the impact of 17 climate change on drought-prone forests is a key objective in assessing forest 18 responses to climate change. 19 Results: In this study, we assessed tree growth trends and changes in physiological 20 activity under climate change based on patterns in tree rings and stable isotopes. 21 Additionally, structural equation models were used to analyze the climate drivers 22 influencing tree growth, with several key results. (1) The climate in the study area 23 showed a trend of warming and drying, with the growth of tree section areas 24 decreasing first and then increasing, while the water use efficiency showed a steady 25 increase. (2) The effects of climate warming on tree growth in the study area have 26 transitioned from negative to positive. The gradual advance of the growing season and 27 the supply of snowmelt water in the early critical period of the growing season are the 28 key factors underlying the reversal of the sensitivity of trees to climate. (3) Variation 29 in water supply has led to different responses of tree growth to warming, and the 30 growth response of Pinus tabuliformis to temperature rise was closely related to 31 increased water availability. 32 Conclusions: Our study indicates that warming is not the cause of forest decline, and 33 instead, drought caused by warming is the main factor causing this change. If 34 adequate water is available during critical periods of the growing season, boreal 35 forests may be better able to withstand rising temperatures and even exhibit increased 36 growth during periods of rising temperatures, forming stronger carbon sinks. However, 37 in semi-arid regions, where water supply is limited, continued warming could lead to 38 reduced forest growth and even death, which would dramatically reduce carbon sinks 39 in arid ecosystems. 40


Introduction 43
Global climate change is putting increasing pressure on forest ecosystems on a 44 global scale (Adams et  The observed decline in productivity was inferred to be mainly related to high 58 temperature and drought stress (Berner et al., 2013;Dulamsuren et al., 2013;59 McDowell, 2011). Diverse studies, ranging from those based on greenhouse 60 experiments (Adams et al., 2009) and forest-climate logic modeling (Williams et al.,61 2013; Williams et al., 2010) to region-scale forest health monitoring (Littell et al.,62 2008; Williams et al., 2013) and global forest reviews of patterns of mortality (Allen 63 homogeneity of the meteorological data. Thus, the meteorological data for these sites 163 were determined to be reliable and without aberrations and thus well represent climate 164 change occurring in the local area. Temperature anomalies and annual precipitation 165 were used to represent climate changes in the study area (Fig. 2), where Temperature 166 Anomalies represent the difference between the annual average Temperature and the 167 multi-year average Temperature. The snowmelt water data were estimated from 168 temperature and winter precipitation data, according to a simple formula (Zhang et  (1) 171 Here, M is the potential snowmelt, in mm/day; C m is the degree-day coefficient, in 172 mm/degree-day; T a is the average daily air temperature (℃); T b is the base 173 temperature (℃). In the calculations used, C m was usually set at 2.74, and T b was set 174 at 0℃. 175 We set up soil temperature and soil volumetric water content measurement 176 probes (Decagon 5TE) at different soil depths (0-20 cm, 20-40 cm, 40-60 cm) in the 177 sample plot, together with an EM50 data collector, to monitor long-term soil 178 temperature and soil volumetric water content since 2018. 179 To quantify the severity of the drought, we used station climate data to estimate 180 standardized precipitation evapotranspiration indices (SPEIs) and saturated vapor 181 pressure difference (VPD) ( Here, R is the radius of the tree and t is the year in which the tree rings were formed. 204 Finally, we calculated the mean BAI chronology for each location (Fig. 3b). The trend 205 of BAI across two consecutive periods (1957-1987 and 1988-2016)  Here, δ 13 C is the stable carbon isotope value of tree-ring cellulose; δ 13 C a is the stable 213 carbon isotope value in atmospheric CO 2 ; Δ is the 13 C discriminant value referring to 214 the difference in isotope levels during photosynthesis between the tree leaf and air; C a 215 represents the concentration of atmospheric CO 2 ; a represents the stomatal 216 fractionation coefficient in the diffusion process, which is about 4.4‰; and b 217 represents the fractionation coefficient in the carboxylation process of Rubisco and 218 PEP carboxylase, which is about 27‰. Additionally, the coefficient 1.6 represents the 219 ratio of the diffusivity of water vapor to CO 2 in the air. The C a values are from 220 NOAA's Earth System Research Laboratory (http://www.esrl.noaa.gov/). 221 We estimated transpiration in the study area from 1957 to 2016 based on annual 222 carbon sequestration and tree iWUE. The annual carbon sequestration estimate was 223 adopted from the biomass model and allometric growth model for Pinus tabuliformis 224 in the study area established by Yang et al. (2021). 225

Statistical analysis 226
We determined the time-dependent relationships between tree growth and 227 climate by using Pearson correlation analysis and Kalman filters. We used a sliding 228 window correlation analysis to assess the change in the correlation coefficient 229 between climate factors and site chronology, and we also calculated the correlation 230 between the temperature of each month during the growing season (March to 231 November) and the growth of trees. The climate is generally considered to be the 232 average of meteorological conditions, e.g., temperature and rainfall, over a 30-year 233 period, as this length of time is considered sufficient to understand the trend in 234 climate change. Accordingly, for our sliding window analyses, we used a fixed 235 window of 30 years, starting with 1957-1986 and ending with 1979-2008, and 236 repeating iterations in one-year increments (Biondi and Waikul, 2004). We also 237 calculated the correlation between tree growth and seasonal (spring, summer, winter) 238 mean climatic variables, as this is more representative of climatic conditions than data 239 from any single month. 240 In order to explore the potential interaction between the water cycle and 241 temperature, we calculated the correlations of temperature with precipitation, 242 evapotranspiration, saturated water pressure difference, and water use efficiency 243 (Littell et al., 2008;Poulter et al., 2013;Repo et al., 2021). We also examined how the 244 interaction between precipitation and growth temperature correlations changed, by 245 using an 11-year time window to assess their consistency over the study period.  1957-1986 and 1987-2016, 251 because the correlation between temperature and water cycle shifted between these 252 two periods. We also examined the relative importance of variables that may affect 253 tree growth using the average BAI of tree rings in the study area, as well as 254 temperature, which was divided into three variables (TEM 3-4 , TEM 5-7 , and TEM 8-10 ). 255 We tested models with different variables, recorded their comparative fit index (CFI), Many studies have suggested that global warming is prone to induce 279 physiological drought, which has a negative effect on tree growth. However, these 280 studies ignore the impact of water supply in the early growing season on tree growth, 281 especially in areas where water is limited. Snowmelt caused by warming has become 282 an important source of water for tree growth in such areas. We analyzed the changes 283 in the growth of tree ring width and sectional area of 10 sample points in this region 284 from 1957 to 2016 and found that the growth of tree ring width and sectional area in 285 this region first increased and then decreased. The inflection point was determined to 286 occur in roughly 1988 by using a Kalman filter approach (Fig. 2). Before 1988, the 287 growth rate of average tree-ring sectional area decreased by 0.799 cm 2 /10a (R = 0.66, 288 P < 0.01). After 1988, the average tree-ring sectional area increased at a rate of 1.799 289 cm 2 /10a (R = 0.43, P < 0.01). This suggests that the response of trees to climate 290 change in this region may have changed around 1988. 291 We found that there was no significant trend in stable carbon isotopes in tree 292 rings in the study area, but the iWUE of Pinus tabuliformis showed a significant 293 increasing trend (Fig. 2c), increasing by 6.68 μmol·mol -1 /10a. The iWUE from 2007 294 to 2017 was 35.7% higher than the average iWUE from 1957 to 1967. In particular, 295 the increasing trend was obviously increased after 1988, reaching 10.29 296 μmol·mol -1 /10a, and this year was consistent with the turning point of the tree growth 297 trend. To some extent, iWUE reflects the degree to which trees are subject to water 298 stress, and higher WUE values indicate more serious water stress. Thus, the growth 299 rate of trees appeared to be faster after 1988. However, owing to the influence of 300 climate warming and drying, the water stress of trees was gradually intensified, 301 leading to a rapid increase in WUE. Year 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996  black solid line represents the measured tree ring data at each sample point, the red 307 solid line represents the average tree rings and BAI variation trend across 10 sample 308 points, and the blue broken line represents the average δ 13 C of tree rings across sites. 309

Impact of climate warming on tree growth 310
We found that the correlation between tree growth and temperature in the study 311 area changed from negative to positive during 1957-2017 (Fig. 4) This change over time was more obvious in the relationship between tree ring 322 growth and seasonal temperature. The correlation between tree ring growth and 323 temperature changed in spring, summer, and autumn ( Fig. 4c and 4d), showing a 324 positive response to temperature. However, for some months, there was also a change 325 in the relationship between tree growth and monthly temperature, while a negative 326 effect of temperature on tree growth was always observed in May, June, and July. The 327 difference in water supply among months may be the main factor underlying this 328 change. 329

Effects of water supply on tree growth throughout the growing season 330
Warmer temperatures may lead to an earlier arrival of the growing season, with 331 the start of the growing season being 1.5 d/10a earlier since 1957 (Fig. 5b). We also 332 found that the end of the growing season began to be delayed at a rate of 1d/10a ( increase, because the soil temperature also changed from negative to positive during 352 the same period (Fig. 6). In the early stage of tree growth, the physiological activities 353 of trees begin to increase their demand for water, and trees are then particularly 354 vulnerable to water stress. We found that the previously observed negative effect of 355 temperature on tree growth in March and April has been reversed since 2000 (Fig. 4) As the influence of temperature on tree growth did not change in the middle of 367 the growing season, we also analyzed the trend in saturated vapor pressure difference 368 (VPD) and evapotranspiration (E T ) in the study area from 1957 to 2017 (Fig. 7). We 369 found that the VPD and E T in the study area increased over time, with VPD increasing 370 by 0.18 hPa/10a and E T by 17.68 mm/10a. All these factors reflect that the climate in 371 the study area is gradually becoming arid, and the water stress on trees is thus 372 gradually becoming intensified. 373 In order to study the relationship between warming and aridity in the study area, 380 we analyzed the trend in Pearson correlation coefficients between temperature and 381 water-related factors over time (i.e., precipitation, VPD, E T , and iWUE) (Fig. 8). Pearson's correlation coefficient over the following three periods of time : 1962-1978, 389 1979-1995, and 1996-2012. 390 We found that the correlation between annual precipitation and temperature was 391 weakly negative overall, with an average correlation coefficient of -0.1. In the 392 warming period, the negative correlation between precipitation and temperature was 393 slightly strengthened (Fig. 8a). There was a positive correlation between temperature 394 and VPD (Fig. 7b), and the correlation increased during the warming period (P < 395 0.01). It can be understood that warming intensifies the water deficit of the 396 atmosphere and to some extent intensifies the climate aridity. The correlation between 397 temperature and E T was also positive (Fig. 7c). However, during the warming period, 398 this correlation weakened (P < 0.01) and gradually became non-significant. This 399 indicates that the effect of temperature on E T becomes weaker as the climate warms, 400 and thus, water supply can gradually become the main factor influencing E T . There 401 was also a positive correlation between temperature and iWUE (Fig. 7d), but the 402 positive correlation gradually weakened (P < 0.01) and even became a negative 403 correlation after 2000. 404 We also found that in the warming period, the correlations between precipitation 405 and saturated water pressure difference, evapotranspiration water use efficiency, and 406 temperature all experienced their own turning points within the study period. The 407 negative effect of temperature on precipitation became reversed around 1978, and it 408 seems that the positive effect of temperature on VPD was also reversed at that time. area exhibited a gradual warming and drying trend (Fig. 2 and 7). Water stress on tree 426 growth gradually intensified (Fig. 3c), and warming may be the main underlying 427 factor. Thus, between May and July, the negative effects of warming on tree growth 428 were not mitigated, and water supply may be the main factor determining the impact 429 of warming on tree growth. 430 To evaluate this hypothesis, we established structural equation models (SEMs) 431 with standard path coefficients for 1957-1986 (Fig. 9a) and 1987-2016 (Fig. 9b) We found that during 1956-1986, the correlation between warming and summer 447 precipitation was low (R = -0.10, P = 0.166), while during 1987-2016, there was a 448 very significant negative correlation between temperature and summer precipitation 449 (R = -0.36, P <0.001) (Fig. 9). SEM analysis also indicated the importance of 450 snowmelt water for tree growth during 1987-2016, as the correlation between average 451 temperature in March and April and tree growth changed between the two periods 452 owing to the influence of snowmelt water. Warming tends to exacerbate drought 453 conditions, and the positive effect of precipitation on tree growth was weakened by 454 the influence of temperature and the compensation effect of SPEI. We also found that 455 the effect of autumn temperature on tree growth was reversed from negative to 456 positive between the two periods, but there was no significant correlation between 457 autumn precipitation and tree growth. Thus, the SEM analysis supports our previous 458 hypothesis that, as a result of climate warming, the growing season of trees is 459 beginning earlier each year in the study area. The growing season has advanced 460 enough to overlap in time with periods of high soil moisture resulting from snowmelt. 461 Thus, the increased supply of snowmelt water has reversed the negative effects of 462 warming on tree growth in the early growing season. However, warming also 463 intensifies drought, and water stress exacerbates the negative effects of warmer 464 summers on tree growth. 465

Effects of water supply in the early and middle stages of tree growing seasons 498
However, it is not immediately obvious why the shift in the temperature-tree 499 growth relationship occurred in March and April. We hypothesize that it is related to 500 the water supply in the early growth period of trees. The growth of Pinus tabuliformis 501 begins in early April (Rossi et al., 2008;Seo et al., 2011). However, the precipitation 502 in our study area is very low in April, averaging only 15.5 mm. We also found that 503 before 1986, the precipitation accumulation was very low in winter and mainly 504 concentrated in summer. The average winter precipitation accumulation during 1956-505 1986 was only 9.1 mm. Combined with the early melting of snow, the water supply to 506 support the growth of wood in the early part of the growing season may be limited 507

Variation in water supply induces different responses of tree growth to 526
warming 527 We found that from May to July, when the temperature-tree growth correlation 528 did not change over the course of the historical data, rainfall showed a significant 529 decreasing trend, while VPD showed a significant increasing trend. However, there 530 was no significant change in rainfall or VPD during the period of August to October, 531 when the temperature-tree growth correlation did indeed change over time. continued warming will not only lead to the decline or even death of forest growth in 566 semi-arid regions, but also lead to the eventual degradation of forests into grasslands 567 in semi-arid regions, which will significantly reduce the total available carbon sink of 568 these altered ecosystems in arid regions. 569 570

Conclusion 571
In this study, we used tree-ring chronology and stable isotopes to assess the 572 response of tree growth to climate change. We found that the effects of climate 573 warming on tree growth transitioned from negative to positive during the period of 574 1957-2017. Adequate water supply during the growing season, especially snowmelt 575 water available in the early part of the growing season, appears to be the key to the 576 reversal of the climate sensitivity of trees in this area. The beginning of the growing 577 season continued to advance gradually throughout the study period. The gradual 578 advance of the beginning of the growing season combined with the availability of 579 snowmelt water early in the growing season resulted in a shift in the response of tree 580 growth to temperature in the study area. Variation in water supply has led to different 581 responses of tree growth to warming throughout the growing seasons. Our study 582 suggests that warming per se is not the direct cause of forest decline, but is indeed the 583 main cause of drought, which generally causes forest decline. SEM analysis also 584 demonstrated that the growth response of Pinus tabuliformis to the observed 585 temperature increase was closely related to the increase in water availability. As a 586 result, boreal forests may be better able to withstand rising temperatures if they have 587 sufficient water, with boosted growth even possible during periods of rising 588 temperatures. However, in semi-arid regions where water supplies are limited, 589 continued warming could lead to reduced forest growth or even death. 590 591