This is the accepted version of the journal article: Wang, Wenjin; Huang, Jian-Guo; Zhang, Tongwen; [et al.]. «Precipitation regulates the responses of xylem phenology of two dominant tree species to temperature in arid and semi-arid forest of the southern Altai Mountains». Science of the total environment, Vol. 886 (Aug. 2023), art. 163951. DOI 10.1016/j.scitotenv.2023.163951 This version is available at https://ddd.uab.cat/record/284784 under the terms of the license 1 Title: Precipitation regulates the responses of xylem phenology of two dominant tree 2 species to temperature in arid and semi-arid forest of the southern Altai Mountains 3 4 Wenjin Wang1,2, Jian-Guo Huang3, Tongwen Zhang4, Li Qin4, Peng Zhou1,2, Shaowei 5 Jiang1, Yaling Zhang1*, Josep Peñuelas5,6 6 1 7 Key Laboratory of Vegetation Restoration and Management of Degraded 8 Ecosystems, South China Botanical Garden, Chinese Academy of Sciences 723 9 Xingke Road, Tianhe District, Guangzhou 510650, China 2 10 University of Chinese Academy of Sciences, 19(A) Yuquan Road, Shijingshan 11 District, Beijing 100049, China 3 12 Key Laboratory of Conservation Biology for Endangered Wildlife of the Ministry of 13 Education, College of Life Sciences, Zhejiang University, Hangzhou, China 4 14 Key Laboratory of Tree-Ring Physical and Chemical Research, CMA/ Xinjiang Key 15 Laboratory of Tree-ring Ecology, Institute of Desert Meteorology, China 16 Meteorological Administration, Urumqi 830002, China 5 17 CSIC, Global Ecology Unit CREAF-CSIC-UAB Bellaterra, Barcelona 08193, 18 Catalonia, Spain 6 19 CREAF, Bellaterra, Barcelona 08193, Catalonia, Spain 20 21 * 22 To whom correspondence may be addressed. E-mail: yaling.zhang@scbg.ac.cn 23 24 1 25 Abstract 26 Arid and semi-arid forests are important carbon (C) sinks, where climate mediated 27 shifts in growing seasons affect biotic interactions, with implications for the global C 28 balance. However, impacts of climate warming on the growth of arid and semi-arid 29 forest tree species and ecosystem C sink dynamics remain uncertain, because of a lack 30 of understanding of effects of the complex interactions between water availability and 31 temperature on phenology. Here, we monitored xylem formation over 2 years in two 32 dominant tree species (Siberian larch, Larix sibirica Ledeb.; Siberian spruce, Picea 33 obovate Ledeb.) with contrasting resource allocation strategies, along the arid and semi- 34 arid southern Altai Mountains of Central Asia. We found that growth of L. sibirica was 35 not limited by early season water availability, as it relies on snow melt for the initiation 36 of spring phenology, while the advanced onset of xylem formation under rising mean 37 annual temperatures was compensated for by earlier cessation of cell enlargement, 38 likely due to legacy effects. In contrast, water limitation, which constrained the onset 39 of xylem formation under rising temperatures in the shade-tolerant P. obovate, appeared 40 to relieve water stress later in the growing season and did not lead to earlier cessation 41 of cell enlargement. Under increasing temperatures, we found that extension of the 42 period of cell wall thickening responsible for biomass production was greater in L. 43 sibirica, whereas extension in the period of cell enlargement, when xylem size increases, 44 was greater in P. obovate. Our results demonstrate that water availability is a key driver 45 of growing season length and C sink dynamics in arid and semi-arid forests under 46 climate warming, due species phenological responses related to their resource 47 allocation strategies. Thus, the integration of tree species resource allocation strategies 48 in climate models will improve estimates of climate-C feedbacks in arid and semi-arid 49 forests under future warming scenarios. 50 51 Keywords: xylem formation; threshold temperature; Phenology; Climate change; 52 lagged effects; end of growing season 2 53 1. Introduction 54 Arid and semi-arid forests are substantial net carbon (C) sinks, characterized by high 55 levels of gross primary productivity and low levels of C losses from respiration 56 (Rotenberg and Yakir, 2010) that contribute to large annual variations in atmospheric 57 carbon dioxide (CO2) (Ahlstrom et al., 2015; Poulter et al., 2014). Under global 58 warming, levels of primary productivity and net CO2 uptake of these types of forest are 59 expected to increase, predominantly due to the extension of growing season length 60 under advancing spring phenology and delayed autumn phenology (He et al., 2015; 61 Keenan et al., 2014). However, the earlier onsets in plant growth under advancing 62 spring phenology have been reported to induce water stress as a legacy effect 63 (Buermann et al., 2013; Zhou et al., 2020) that may limit tree growth in cold and dry 64 landscapes (Gao et al., 2022). For example, warming-induced intensified drought 65 conditions (Allen et al., 2010; Wu et al., 2022) have resulted in widespread declines 66 and mortality of arid and semi-arid forests in the western United States (Stanke et al., 67 2021), central Asia (Liu et al., 2013), north-eastern Tibetan plateau (Liang et al., 2015), 68 and eastern Mediterranean (Sarris et al., 2007), although understanding of the combined 69 effects of reduced water availability and rising temperatures on tree growth lacks clarity. 70 71 While xylem formation (xylogenesis) in tree stems is essential for the sequestration and 72 storage of atmospheric C (Xue et al., 2017), the process is sensitive to climate change 73 and exhibits high levels of response plasticity to changing environmental conditions 74 (Fonti et al., 2010; Rossi et al., 2014). Temperature drives spring phenology of 75 xylogenesis (Huang et al., 2020; Huang et al., 2022), including along altitudinal 76 gradients (Moser et al., 2010; Saderi et al., 2019); however, understanding of combined 77 impacts of shifts in temperature and water availability (Forkel et al., 2015; Garonna et 78 al., 2018) under climate change on xylogenesis phenology in arid and semi-arid areas 79 is unclear. For example, although studies have shown a mean stable air temperature 80 threshold of 5.6–8.5 °C for the triggering of xylogenesis in alpine tree species (Rossi et 81 al., 2007; Rossi et al., 2008), there is evidence for water availability as a principal driver 82 of spring xylem formation in some arid and semi-arid areas (Ren et al., 2015; Ziaco et 3 83 al., 2018). 84 85 Earlier spring phenology may trigger an advancement in autumn phenology, due to the 86 associated lagged effect of accelerated levels of growing season water stress (Buermann 87 et al., 2018). While this role of lagged effects on timing of autumn phenology of impacts 88 of climate warming on water availability in primary growth is understood (Keenan and 89 Richardson, 2015; Zani et al., 2020), it remains unknown for secondary growth, 90 including xylem phenology, and contributes to the uncertainty of drivers of xylogenesis. 91 For example, the lack of association between cessation of xylogenesis and elevation 92 reported for the north-western Himalayas (Malik et al., 2020), north-eastern Tibetan 93 Plateau (Zhang et al., 2018b), and southern European Alps (Moser et al., 2010; Saderi 94 et al., 2019), indicate a lack of influence of temperature on termination of xylem cell 95 differentiation and a likely impact of water limitation, given the clear role of soil water 96 potential in cambium activity and xylem formation (Cabon et al., 2020; Eckes- 97 Shephard et al., 2021) and effects of reduced soil water availability on advancement in 98 cessation of xylem growth (Miller et al., 2022; Moser et al., 2010). Thus, there is a lack 99 of understanding of the combined effects of shifts in temperature and water availability 100 on the ending of xylem formation that restricts confidence in predictions of potential 101 effects of climate change on forest growth. 102 103 The Altai Mountain range in arid and semi-arid central Asia extends for more than 2000 104 km along a southeast-northwest orientation and acts as a climate boundary, where the 105 convergence of westerly airflows with monsoon airflows from the east (Huang et al., 106 2018; Zhang et al., 2018a) results in a gradient in mean annual precipitation along the 107 range, from 400–700 mm in the northwest to less than 200 mm in the southeast (Huang 108 et al., 2018). Analyses of tree rings across the region have revealed temporally unstable 109 and spatially heterogeneous responses of forest growth to climate change (Jiang et al., 110 2021; Taynik et al., 2016), including elevational variations in forest growth responses 111 to temperature and precipitation in the southern Altai Mountains (Zhou et al., 2021); 112 however, the combined impacts of water availability and temperature on xylogenesis in 4 113 arid and semi-arid forest tree species and implications for forest growth and C storage 114 are unclear. 115 116 Here, we conducted a two-year field study in the southern Altai Mountains of cambium 117 activity and xylem formation in two dominant tree species, Siberian larch (Larix 118 sibirica Ledeb.) and Siberian spruce (Picea obovate Ledeb.) to test for 1) combined 119 effects of growing season water availability and temperature on the timing of xylem 120 phenology, and 2) lagged effects of earlier onset of late growing season xylogenesis. 121 The overarching aim of our study was to test the hypotheses that there are high levels 122 of variability in xylem spring and autumn phenology in response to water availability 123 and temperature and that earlier spring xylem phenology leads to lagged effects on tree 124 growth in the arid and semi-arid Altai Mountains. 125 126 2. Materials and methods 127 2.1 Study sites 128 The study sites were located in the southern Altai Mountains, where there is a prevailing 129 cold continental climate, with a mean annual temperature of c. 2.3 °C and mean annual 130 precipitation of 254.2 mm over the period 1963 to 2016 (Zhou et al., 2021). We selected 131 four forest stands distributed at contrasting altitudes along c. 400-km of the mountain 132 range comprising Altay low altitude (AL; 47°5736N, 88°60E) at 1240 m a.s.l., 133 Kanas (KN; 48°4312 N, 87°148E) at 1437 m a.s.l., Qinghe (QH; 46°5736N, 134 90°1424E) at 1589 m a.s.l., and Altay high altitude (AH; 48°036N, 88°1836E) at 135 1829 m a.s.l. (Figure S1). 136 137 2.2 Tree sampling and analysis 138 In April 2018, we selected 70 healthy L. sibirica and P. obovate individuals that 139 represented the dominant species at the four study sites, comprising 10 of each species 140 at AL, KN, and QH, and five of each species at AH, due to lack of suitable individuals 141 (Table S1); in 2019, five P. obovate at AL were lost, due to a snowstorm. 142 5 143 Each week, from April to October of 2019 and 2020, we used a Trephor (Rossi et al., 144 2006a) to collect 2 mm × 1.5 cm microcore samples of phloem, cambium zone, and 145 forming and previously formed rings from the tree stems at 1.3 m above ground level 146 (at breast height). . The microcore samples were placed in microtubes containing 50% 147 ethanol and stored at 5 °C prior to analysis of xylem formation, when they were 148 dehydrated using successive immersions in ethanol and D-limonene and embedded in 149 paraffin blocks. We cut the microcores using a rotary microtome to obtain 8–10 μm 150 transverse sections that were then stained with cresyl violet acetate (0.06%) and 151 examined under light microscopy (bright-field and polarized light) at ×40–400 152 magnification (Rossi et al., 2006b). We extracted c. 4,600 transverse sections of 153 microcore from the 70 trees over the study period. 154 155 We randomly selected three rows of radial cells from each section and counted the 156 number of cambium cells, enlarging cells, wall-thickening cells, and mature cells, based 157 on distinct morphological characteristics (Rossi et al., 2006b, 2012). 158 159 2.3 Meteorological data 160 A meteorological station was installed at each study site in March 2019 to record daily 161 minimum, maximum, and mean air temperatures; missing temperature data were 162 substituted from nearby weather stations. We extracted precipitation data from the GPM 163 IMERG V06 dataset (https://www.ncdc.noaa.gov) (Huffman et al., 2019), which 164 estimates and detects regional spatial and temporal precipitation patterns (Pradhan et 165 al., 2022), and compared these data with monthly precipitation data collected at three 166 national meteorological stations distributed along the southern Altai mountains for the 167 period 2017–2019; we found an association between the two datasets (R=0.75), 168 indicating that the GPM IMERG V06 dataset reflects seasonal patterns in precipitation 169 across the study sites. 170 171 2.4 Metrics of xylem phenology 172 For the following xylem phenology metrics, we recorded day of year (DOY) dates of 6 173 (1) first enlarging cell (bE); (2) first wall-thickening cell (bW); (4) last enlarging cell 174 (cE); and, (5) last wall-thickening cell (cW). The duration of cell enlargement (dE), 175 duration of cell wall thickening (dW), and duration of xylem formation (dX) are 176 phenological events was computed as the difference between cE and bE, cW and bW, 177 and cW and bE, respectively (Rossi et al., 2012). 178 179 2.5 Estimates of phenology threshold temperature 180 The probability of phenological events at a given air temperature was estimated using 181 logistic regression, where binary variables were based the presence or absence of the 182 observed event; for example, the presence and absence of enlarging cells were coded 183 as 1 and 0, respectively. The model was fitted with mean air temperature for each tree, 184 species, site, and year as: 𝜋𝑥 185 Logit(𝜋𝑥 ) = ln ( ) = 𝛽0 + 𝛽1 𝑥𝑗 1 − 𝜋𝑥 186 where 𝜋𝑥 is the probability of phenological events occurring; 𝑥𝑗 is the temperature 187 on day j; and, 𝛽0 and 𝛽1 are the intercept and slope, respectively (Rossi et al., 2007; 188 Rossi et al., 2008). Temperature thresholds (x) were calculated when the probability of 189 phenological events occurring was 0.5, where Logit(𝜋𝑥 ) = 0 and x = − 𝛽0⁄𝛽1; thus, 190 xylem phenological events were more likely to occur at temperatures >x. Model fit was 191 checked using 𝜒 2 of the likelihood ratio, Wald’s 𝜒 2 for regression parameters and 192 goodness of fit, and Hosmer–Lemeshow Ĉ for eventual lack of fit (Quinn and Keough, 193 2002). 194 195 2.6 Statistical analyses 196 Species differences in timings of xylem phenology were tested using a bootstrap student 197 t-test, resampled 1000 times, while study site and year differences for each species were 198 estimated using generalized linear mixed models (GLMMs), with site and year as fixed 199 effects and tree as a random effect. We tested for effects of mean annual temperature 200 (MAT) and total annual precipitation (TAP) on xylem phenology timing for each 201 species using GLMMs, with MAT and TAP as fixed effects, and tree nested within study 7 202 site as a random effect. GLMMs allow the non-normal distribution of errors with the 203 utilization of a link function; therefore, we used the log function, given xylem 204 phenology timings are positive integers. 205 206 Species differences in phenology threshold temperatures were tested using a bootstrap 207 student t-test, resampled 1000 times, while site and year differences were tested using 208 two-way ANOVA with a permutation test, given these data were non-normally 209 distributed (Anderson and ter Braak, 2010). We used linear mixed models (LMMs) to 210 estimate effects of MAT and TAP on threshold temperatures for each species, with MAT 211 and TAP as fixed effects and tree nested in study site as a random effect. We tested for 212 assumptions of normality and homoscedasticity of model residuals. We tested for 213 interactions between effects of MAT on threshold temperatures and TAP using simple- 214 effects analysis, based on TAP values of 164 mm (minimum value at QH in 2019), 184 215 mm (mean TAP minus SD), 267 mm (mean TAP minus SD), 349 mm (mean TAP plus 216 SD), 388 mm (maximum value at KN in 2019). 217 218 Data analysis and modeling were conducted using R (R Core Team, 2022, version 4.2.1), 219 using packages “CAVIAR” (Rathgeber et al., 2018), “lme4” (Bates et al., 2014), 220 “lmPerm” (Wheeler and Torchiano, 2016), and “effects” (Fox and Weisberg, 2019). 221 222 3. Results 223 3.1 Study site climate conditions 224 Across the four study sites, MAT was lowest at AH and highest at AL and at QH, MAT 225 was higher in 2019 than 2018 (Figure 1a), while TAP decreased from the northwest 226 (KN) to southeast (QH) (Figure 1b). 227 8 228 229 Figure 1. Mean (±SE) annual temperature (a) and total annual precipitation (b) at the 230 four monitoring sites in the southern Altai Mountains in 2018 and 2019. KN: Kanas; 231 AL: Altai low altitude; AH: Altai high altitude; QH: Qinghe. 232 233 3.2 Threshold temperatures for onset of xylogenesis 234 There were year and study site differences in threshold temperatures for the onset of 235 xylogenesis in L. sibirica (P < 0.001; Table S2), where thresholds were higher in 2018 236 and ranged from 7.5 °C at QH in 2019 to 12.6 °C at AL in 2018 (Figure 2a). For P. 237 obovate, there were site differences in threshold temperatures, where thresholds ranged 238 from 7.5 °C at KN in 2018 to 11.2 °C at AL in and 2018 and 2019 (Figure 2b); effects 239 of site on threshold temperature varied between year (P < 0.001; Table S2). Across 240 study sites, there were species differences in mean threshold temperatures for the onset 241 of xylogenesis (P = 0.042), where there was a higher threshold for L. sibirica (10.0 °C) 242 than for P. obovate (9.3 °C) 243 9 244 245 Figure 2. Study site estimates of mean threshold air temperatures corresponding with 246 the 0.5-probability of onset of xylogenesis for L. sibirica and P. obovate for 2018 and 247 2019. KN: Kanas; AL: Altai low altitude; AH: Altai high altitude; QH: Qinghe; error 248 bars are SE. 249 250 Analysis showed effects of negative MAT on threshold temperatures for the onset of 251 xylogenesis in L. sibirica (P < 0.012) and P. obovate (P < 0.001) and variation in effects 252 of TAP on impacts of MAT on threshold temperatures (P < 0.026 and P < 0.001, 253 respectively) (Table S3). For L. sibirica, the negative effects of MAT on threshold 254 temperatures decreased with an increase in TAP (Figure 3a), while for P. obovate, 255 threshold temperatures decreased with increasing MAT when TAP is <335mm and the 256 reverse when TAP is >335mm (Figure 3b). 257 258 10 259 260 Figure 3. Relationship between threshold temperatures for the onset of xylogenesis and 261 mean annual temperature (MAT) under contrasting levels of total annual precipitation 262 (TAP) recorded at the four study sites for L. sibirica (a) and P. obovate (b). Vertical and 263 horizontal dashed lines represent MAT of 1.6 C and xylogenesis onset threshold 264 temperature of 6.7 C at 335 mm TAP. 265 266 3.3 Xylem phenology 267 Across the study sites, onset of cell enlargement and wall thickening tended to occur in 268 late May and mid-June, respectively, and tended to be earlier AL (Figure 4). Onset 269 timings of cell enlargement and wall thickening varied in L. sibirica between study sites 270 (P = 0.001 and P = 0.005, respectively) and years (P = 0.013 and P = 0.003, respectively) 271 and onset of cell enlargement in P. obovate varied between study sites (P = 0.006) 272 (Table 1) 273 274 Across the study sites, cessation of cell enlargement in L. sibirica occurred c. mid-July 275 and c. early August in P. obovate, and cell wall thickening ceased c. mid-August and 276 early September, respectively; cell enlargement tended to cease earlier at AL and KN 277 (c. mid-July) than at AH and QH (late July to mid-August) (Figure 4). Timing of 278 cessation of cell enlargement in L. sibirica varied with site (P = 0.001) and year (P = 279 0.041) (Table 1) and was earlier than in P. obovate (P < 0.05), leading to a shorter period 11 280 of xylem cell enlargement, wall thickening, and xylem growing season (P < 0.05) 281 (Table S4). 282 283 284 Figure 4. Mean (±SE) duration of cell enlargement (E) and wall thickening (W) in L. 285 sibirica and P. obovate at the four study sites in 2018 and 2019. DOY: day of the year; 286 KN: Kanas; AL: Altai low altitude; AH: Altai high altitude; QH: Qinghe. 287 288 Table 1. Generalized linear mixed effects model estimates of effects of study site (S) 289 and year (Y) on phenology of cell enlargement and cell wall thickening in L. sibirica 290 and P. obovate. Cell enlargement Cell wall thickening Phenology Fixed metric effects L. sibirica P. obovate L. sibirica P. obovate χ2 P χ2 P χ2 P χ2 P Onset S 15.947 0.001 12.435 0.006 12.749 0.005 3.716 0.294 Y 6.169 0.013 0.205 0.651 8.950 0.003 0.001 0.975 S×Y 0.279 0.963 1.002 0.801 2.974 0.396 0.996 0.802 Cessation S 21.075 0.000 2.053 0.561 4.283 0.233 2.913 0.405 Y 4.156 0.041 0.014 0.906 0.052 0.820 0.002 0.961 S×Y 3.755 0.289 4.438 0.218 1.598 0.660 3.350 0.341 291 P-values in bold are <0.05. 12 292 293 3.4 Climate drivers of xylem phenology 294 We found species differences in effects of MAT and TAP on onset timings of xylem cell 295 development, where timings of onset of cell enlargement and wall thickening in L. 296 sibirica were negatively affected by MAT and positively affected by TAP (Figure 5a, 297 c), while in P. obovate, onset timing of cell enlargement was negatively impacted by 298 MAT (Figure 5b) (Table S5). 299 300 There were few effects of MAT and TAP on cessation of xylem phenology in the tree 301 species: timing of cessation of cell enlargement in L. sibirica was negatively influenced 302 by MAT and effects of MAT varied with TAP (Figure 5d; Table S6). 303 304 305 306 Figure 5. Generalized linear mixed model estimates of effects of mean annual 307 temperature (MAT) and total annual precipitation (TAP) on variation in the onset timing 13 308 of cell enlargement (DOY) in L. sibirica (a) and P. obovate (b), onset timing of cell wall 309 thickening (DOY) in L. sibirica (c), and cessation timing of cell enlargement (DOY) in 310 L. sibirica (d). 311 312 4. Discussion 313 4.1 Climate drivers of onset of xylogenesis 314 We found a large range in threshold temperatures (c. 5 °C) for onset of xylogenesis in 315 L. sibirica and P. obovate, where there were differences among study sites and between 316 years (Figure 2; Table S2), indicating a lack of stable threshold temperature in the arid 317 and semi-arid southern Altai Mountains, supporting findings from the arid and semi- 318 arid northeastern Tibetan Plateau (Ren et al., 2018), but in contrast to studies from 319 European Alpine treelines, where conditions are wetter (Rossi et al., 2007; Rossi et al., 320 2008). This absence of a common threshold temperature in these relatively dry 321 conditions is probably due to effects of early growing season interactions between 322 temperature and water availability, given that, in addition to temperature, preseason soil 323 water conditions, due to snowmelt (Castagneri et al., 2018) and precipitation (Shamir 324 et al., 2020), drive the resumption of tree growth following a period of rehydration in 325 spring to compensate for water lost during winter and ensure an adequate water balance 326 (Turcotte et al., 2009) to support cell division and expansion. 327 328 Our results revealed species differences in the role of TAP in the regulation of the 329 relationship between threshold temperatures and MAT. We found negative effects of 330 MAT on the onset timings of cell enlargement in L. sibirica that were weakened with 331 increasing TAP (Figure 3a), indicating that cell enlargement in response to rising 332 temperatures was not constrained by water limitation and that climate warming may 333 advance the onset of cell enlargement. Our finding of delayed effects of TAP on onset 334 times of cell enlargement and wall thickening (Figure 5a) supported previous research 335 in wetter areas (Shen et al., 2015), but were in contrast to findings of studies in arid and 336 semi-arid areas (Ren et al., 2015; Ziaco et al., 2018), possibly because high levels of 337 precipitation tend to coincide with low levels of insolation intensity and duration that 14 338 delay xylogenesis onset, due to minimal heat requirements (Shen et al., 2015), as 339 indicated by the positive association between growing degree days required for the start 340 of the vegetation growing season and preseason precipitation across the Tibetan Plateau 341 (Fu et al., 2014). 342 343 In contrast for P. obovate, we found optimal water conditions (TAP of c. 335 mm) for 344 the initiation of xylem growth in our study area. Similarly, Ren et al. (2018) reported 345 both a 12-day average minimum temperature of 1.5 ± 1.4 °C and a cumulative 346 precipitation threshold of 17.0 ±5.6 mm required for the onset of xylogenesis in 347 Juniperus przewalskii in the arid and semi-arid north-eastern Tibetan Plateau, while 348 critical soil water potential thresholds have been demonstrated for radial growth (Eckes- 349 Shephard et al., 2021; Peters et al., 2021). Given we found that TAP of the four study 350 sites tended to be below the threshold of 335 mm (with the exception of KN), the onset 351 of xylogenesis in P. obovate is likely to be sensitive to water stress. 352 353 The contrasting species responses of xylem growth onset to precipitation, where high 354 levels of precipitation adversely affect the onset of xylogenesis in L. sibirica, but not in 355 P. obovate that may be sensitive to water stress, may derive from differences in 356 physiological and ecological traits. For example, L. sibirica is shade-sensitive, and 357 adapted to more exposed, dry conditions, whereas P. obovate is shade-tolerant and is 358 adapted for growth in lower light and more humid conditions (Jiang et al., 2014), and 359 the higher levels of stomatal conductance and plastic response to temperature in L. 360 sibirica lead to its greater water use efficiency than P. obovate (Peters et al., 2019); it is 361 possible that water in the arid and semi-arid southern Altai Mountains needed for L. 362 sibirica growth (Jiang et al., 2021) is provided by the winter snowmelt water, as 363 demonstrated elsewhere (Zhang et al., 2021b). 364 365 While there was a lack of water limitation in growth of L. sibirica, it was sensitive to 366 rising temperatures, as indicated by study site and year differences in onset timings of 367 cell enlargement and wall thickening(Table 1), as sites were proxies of temperature 15 368 (Figure 1a), where earliest onset of xylem cell differentiation was recorded at the 369 warmest study site (AL) in the warmer of the two study years (2019). Our results are 370 consistent with previous studies that report a positive link between earlier start of xylem 371 formation and temperature along elevational or latitudinal gradients (Moser et al., 2010; 372 Rossi et al., 2014), likely because warmer temperatures induce an earlier achievement 373 of thermal accumulation required for cambium activity (Huang et al., 2020; Huang et 374 al., 2022; Rossi et al., 2014) that breaks ecodormancy as environmental conditions 375 become favorable and triggers cell division and enlargement (Begum et al., 2017). 376 377 Sensitivity to drought and lower water use efficiency in P. obovate lead to greater levels 378 of sensitivity to water deficit than L. sibirica, where insufficient water availability may 379 decrease the sensitivity of xylem spring phenology to temperature, given Cong et al. 380 (2013) reported that the sensitivity of spring green-up onset date to temperature 381 increased with increasing precipitation over temperate China. We found evidence for 382 impacts of water limitation on effects of temperature on the onset of xylem cell 383 differentiation in P. obovate, because there were only study site differences in the onset 384 of cell enlargement, rather than year differences, likely because there were smaller 385 differences in temperature between years than among study sites, and we found a 386 positive effect of MAT on onset timing of cell wall thickening in L. sibirica , but not P. 387 obovate, possibly due to legacy effects of water limitation. Previous studies have shown 388 that warming-induced, earlier start to growing seasons may lead to advanced autumn 389 senescence, through increased water stress (Keenan and Richardson, 2015); however, 390 our observations indicate this legacy effect in P. obovate probably occurred in the 391 subsequent spring phenology event of cell wall thickening that may be more sensitive 392 to water availability than cell enlargement, and thus showed no response to MAT. 393 394 4.2 Climate drivers of cessation of xylogenesis 395 We observed no impacts of MAT or TAP on the cessation timings of cell wall thickening 396 in the two tree species, with no variation among study sites or between years, supporting 397 findings of a study from semi-arid mountain forest ecosystems along altitudinal 16 398 gradients (Zhang et al., 2018b), where the termination of cell wall thickening showed a 399 minor correlation with elevation, indicating a minimal influence of temperature on the 400 end of xylogenesis. However, our findings, which are inconsistent with those of Zhang 401 et al. (2021a), who reported a positive correlation between the end of xylogenesis and 402 TAP, imply that precipitation was not a limiting factor for the completion of cell wall 403 lignification and deposition in the southern Altai Mountains. Cell wall thickening 404 involves complex biosynthesis and deposition of cellulose, hemicellulose, cell wall 405 proteins, and lignin (Plomion et al., 2001) that are mainly regulated by C supply 406 (Deslauriers et al., 2016; Verbancic et al., 2018). Thus, climate variables, other than 407 temperature and precipitation, that are associated with photosynthate limitation may 408 also control the course of cell wall thickening, such as solar radiation, photoperiod (De 409 Micco et al., 2019; Jyske et al., 2014), nutrient limitation, and soil water moisture 410 (Miller et al., 2022). 411 412 In contrast, we found that cessation timings of cell enlargement varied among sites and 413 between years and increases in MAT led to earlier cessation timing in L. sibirica, but 414 not P. obovate; we suggest two possible explanations for these findings. Firstly, cell 415 enlargement terminated from mid-July to early August when daily air temperatures 416 peaked and water availability was limited, and the greater density of fine roots in the 417 top soil developed by P. obovate may lead to greater plasticity of water uptake under 418 variable levels of soil moisture (Zhang et al., 2021b) that buffer against the adverse 419 effects of relatively dry conditions during the mid-growing season. Secondly, there may 420 be differences in time lags of legacy effect between the two species, due to plant eco- 421 hydrological properties, such as rooting system and stomatal conductance (Wu et al., 422 2018); for example, the legacy effect of earlier onset of spring phenology may advance 423 cessation of autumn phenology in L. sibirica, due to intensified growth activity (Zani 424 et al., 2020), or water limitation (Buermann et al., 2013; Wolf et al., 2016). 425 426 5. Conclusion 427 17 428 Our results demonstrated that water availability plays a fundamental and complex role 429 in the regulation of xylem phenological responses to rising temperatures in dominant 430 tree species of arid and semi-arid forests, due to plant water stress adaptation strategies 431 (Wu et al., 2018). We found that as L. sibirica relies on snow melt for the initiation of 432 spring growth, phenological onset dates were not constrained by early growing season 433 water limitation, whereas high levels of precipitation and low temperatures associated 434 with low mean insolation intensity and duration in the early growing season led to 435 delayed onset of cell enlargement and wall thickening; earlier onsets of spring xylem 436 phenology in L. sibirica led to earlier cessation of cell enlargement, likely due to 437 increasing levels of water stress through the growing season (Bigler and Vitasse, 2021; 438 Buermann et al., 2018). In contrast, early growing season water limitation in shade- 439 tolerant P. obovate constrained onsets of xylem phenology to rising temperatures, but 440 appeared not to drive water stress later in the season, due to legacy effects, that did not 441 lead to earlier cessation of cell enlargement. The growing period of cell wall thickening, 442 which accounts for 90% of woody biomass production, was longer in L. sibirica, 443 whereas the period of cell enlargement, when cells are produced and grow (Cuny et al., 444 2015), was greater in P. obovate. These contrasting phenological responses reflect 445 species differences in resource allocation strategies, where L. sibirica tends to invest 446 greater amounts of carbohydrate to biomass production under rising temperatures, 447 while P. obovate tends to increase xylem size; thus, phenological responses of these 448 species to climate warming has implications for structure and function of arid and semi- 449 arid forest ecosystems, along with uncertainty of future C sink dynamics (Poulter et al., 450 2014). 451 452 Acknowledgements 453 The authors thank the Xingjiang Kanas National Nature Reserve and Altai Forestry 454 Administrative Bureau for permission to establish the study sites. This work was 455 supported by the Xinjiang Regional Collaborative Innovation Project (2022E01045), 456 the National Natural Science Foundation of China (grant numbers 32271653, 457 32001138), and the Zhejiang University (108000*1942222R1). J.P. was supported by 18 458 the Fundación Ramón Areces Project CIVP20A6621, and the Spanish Government 459 projects PID2019-110521GB-I00 and TED2021-132627B-I00 funded by MCIN, 460 AEI/10.13039/501100011033, and the European Union Next Generation EU/PRTR). 461 References 462 Ahlstrom A, Raupach MR, Schurgers G, Smith B, Arneth A, Jung M, et al. The dominant role of semi-arid 463 ecosystems in the trend and variability of the land CO2 sink. Science 2015; 348: 895-899. 464 Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier M, et al. A global overview 465 of drought and heat-induced tree mortality reveals emerging climate change risks for forests. 466 Forest Ecology and Management 2010; 259: 660-684. 467 Anderson M, ter Braak C. Permutation tests for multi-factorial analysis of variance. Journal of Statistical 468 Computation and Simulation 2010; 73: 85-113. 469 Bates D, Mächler M, Bolker B, Walker S. Fitting Linear Mixed-Effects Models Using lme4. ArXiv e-prints 470 2014; arXiv:1406. 471 Begum S, Kudo K, Rahman MH, Nakaba S, Yamagishi Y, Nabeshima E, et al. Climate change and the 472 regulation of wood formation in trees by temperature. Trees 2017; 32: 3-15. 473 Bigler C, Vitasse Y. Premature leaf discoloration of European deciduous trees is caused by drought and 474 heat in late spring and cold spells in early fall. Agricultural and Forest Meteorology 2021; 307. 475 Buermann W, Bikash PR, Jung M, Burn DH, Reichstein M. Earlier springs decrease peak summer 476 productivity in North American boreal forests. Environmental Research Letters 2013; 8. 477 Buermann W, Forkel M, O'Sullivan M, Sitch S, Friedlingstein P, Haverd V, et al. Widespread seasonal 478 compensation effects of spring warming on northern plant productivity. Nature 2018; 562: 479 110-114. 480 Cabon A, Fernandez-de-Una L, Gea-Izquierdo G, Meinzer FC, Woodruff DR, Martinez-Vilalta J, et al. 481 Water potential control of turgor-driven tracheid enlargement in Scots pine at its xeric 482 distribution edge. New Phytol 2020; 225: 209-221. 483 Castagneri D, Battipaglia G, von Arx G, Pacheco A, Carrer M. Tree-ring anatomy and carbon isotope ratio 484 show both direct and legacy effects of climate on bimodal xylem formation in Pinus pinea. Tree 485 Physiol 2018; 38: 1098-1109. 486 Cong N, Wang T, Nan H, Ma Y, Wang X, Myneni RB, et al. Changes in satellite-derived spring vegetation 487 green-up date and its linkage to climate in China from 1982 to 2010: a multimethod analysis. 488 Glob Chang Biol 2013; 19: 881-91. 489 Cuny HE, Rathgeber CB, Frank D, Fonti P, Makinen H, Prislan P, et al. Woody biomass production lags 490 stem-girth increase by over one month in coniferous forests. Nat Plants 2015; 1: 15160. 491 De Micco V, Carrer M, Rathgeber CBK, Julio Camarero J, Voltas J, Cherubini P, et al. From xylogenesis to 492 tree rings: wood traits to investigate tree response to environmental changes. IAWA Journal 493 2019; 40: 155-182. 494 Deslauriers A, Huang JG, Balducci L, Beaulieu M, Rossi S. The Contribution of Carbon and Water in 495 Modulating Wood Formation in Black Spruce Saplings. Plant Physiol 2016; 170: 2072-84. 496 Eckes-Shephard AH, Tiavlovsky E, Chen Y, Fonti P, Friend AD. Direct response of tree growth to soil water 497 and its implications for terrestrial carbon cycle modelling. Glob Chang Biol 2021; 27: 121-135. 498 Fonti P, von Arx G, Garcia-Gonzalez I, Eilmann B, Sass-Klaassen U, Gartner H, et al. Studying global 499 change through investigation of the plastic responses of xylem anatomy in tree rings. New 19 500 Phytol 2010; 185: 42-53. 501 Forkel M, Migliavacca M, Thonicke K, Reichstein M, Schaphoff S, Weber U, et al. Codominant water 502 control on global interannual variability and trends in land surface phenology and greenness. 503 Glob Chang Biol 2015; 21: 3414-35. 504 Fu YH, Piao S, Zhao H, Jeong SJ, Wang X, Vitasse Y, et al. Unexpected role of winter precipitation in 505 determining heat requirement for spring vegetation green-up at northern middle and high 506 latitudes. Glob Chang Biol 2014; 20: 3743-55. 507 Gao S, Liang E, Liu R, Babst F, Camarero JJ, Fu YH, et al. An earlier start of the thermal growing season 508 enhances tree growth in cold humid areas but not in dry areas. Nat Ecol Evol 2022; 6: 397-404. 509 Garonna I, de Jong R, Stöckli R, Schmid B, Schenkel D, Schimel D, et al. Shifting relative importance of 510 climatic constraints on land surface phenology. Environmental Research Letters 2018; 13. 511 He Z, Du J, Zhao W, Yang J, Chen L, Zhu X, et al. Assessing temperature sensitivity of subalpine shrub 512 phenology in semi-arid mountain regions of China. Agricultural and Forest Meteorology 2015; 513 213: 42-52. 514 Huang JG, Ma Q, Rossi S, Biondi F, Deslauriers A, Fonti P, et al. Photoperiod and temperature as 515 dominant environmental drivers triggering secondary growth resumption in Northern 516 Hemisphere conifers. Proc Natl Acad Sci U S A 2020; 117: 20645-20652. 517 Huang JG, Zhang Y, Wang M, Yu X, Deslauriers A, Fonti P, et al. A critical thermal transition driving spring 518 phenology of Northern Hemisphere conifers. Glob Chang Biol 2022. 519 Huang X, Peng W, Rudaya N, Grimm EC, Chen X, Cao X, et al. Holocene Vegetation and Climate Dynamics 520 in the Altai Mountains and Surrounding Areas. Geophysical Research Letters 2018; 45: 6628- 521 6636. 522 Jiang S, Liang H, Zhou P, Wang Z, Zhu H, Kang J, et al. Spatial and temporal differences in the response 523 of Larix sibirica to climate change in the central Altai Mountains. Dendrochronologia 2021; 67. 524 Jiang Y, Zhang W, Wang M, Kang M, Dong M. Radial growth of two dominant montane conifer tree 525 species in response to climate change in North-Central China. PLoS One 2014; 9: e112537. 526 Jyske T, Mäkinen H, Kalliokoski T, Nöjd P. Intra-annual tracheid production of Norway spruce and Scots 527 pine across a latitudinal gradient in Finland. Agricultural and Forest Meteorology 2014; 194: 528 241-254. 529 Keenan TF, Gray J, Friedl MA, Toomey M, Bohrer G, Hollinger DY, et al. Net carbon uptake has increased 530 through warming-induced changes in temperate forest phenology. Nature Climate Change 531 2014; 4: 598-604. 532 Keenan TF, Richardson AD. The timing of autumn senescence is affected by the timing of spring 533 phenology: implications for predictive models. Glob Chang Biol 2015; 21: 2634-2641. 534 Liang E, Leuschner C, Dulamsuren C, Wagner B, Hauck M. Global warming-related tree growth decline 535 and mortality on the north-eastern Tibetan plateau. Climatic Change 2015; 134: 163-176. 536 Liu H, Park Williams A, Allen CD, Guo D, Wu X, Anenkhonov OA, et al. Rapid warming accelerates tree 537 growth decline in semi-arid forests of Inner Asia. Glob Chang Biol 2013; 19: 2500-10. 538 Malik R, Rossi S, Sukumar R. Variations in the timing of different phenological stages of cambial activity 539 in Abies pindrow (Royle) along an elevation gradient in the north-western Himalaya. 540 Dendrochronologia 2020; 59. 541 Miller TW, Stangler DF, Larysch E, Honer H, Seifert T, Puhlmann H, et al. Longer and faster: Intra-annual 542 growth dynamics of Douglas fir outperform Norway spruce and silver fir over wide climatic 543 gradients. Agricultural and Forest Meteorology 2022; 321. 20 544 Moser L, Fonti P, Buntgen U, Esper J, Luterbacher J, Franzen J, et al. Timing and duration of European 545 larch growing season along altitudinal gradients in the Swiss Alps. Tree Physiol 2010; 30: 225- 546 33. 547 Peters RL, Speich M, Pappas C, Kahmen A, von Arx G, Graf Pannatier E, et al. Contrasting stomatal 548 sensitivity to temperature and soil drought in mature alpine conifers. Plant Cell Environ 2019; 549 42: 1674-1689. 550 Peters RL, Steppe K, Cuny HE, De Pauw DJW, Frank DC, Schaub M, et al. Turgor - a limiting factor for 551 radial growth in mature conifers along an elevational gradient. New Phytol 2021; 229: 213-229. 552 Plomion C, Leprovost Gg, Stokes A. Wood Formation in Trees. Plant Physiology 2001; 127: 1513-1523. 553 Poulter B, Frank D, Ciais P, Myneni RB, Andela N, Bi J, et al. Contribution of semi-arid ecosystems to 554 interannual variability of the global carbon cycle. Nature 2014; 509: 600-3. 555 Pradhan RK, Markonis Y, Vargas Godoy MR, Villalba-Pradas A, Andreadis KM, Nikolopoulos EI, et al. 556 Review of GPM IMERG performance: A global perspective. Remote Sensing of Environment 557 2022; 268. 558 Quinn G, Keough M. Experimental Design and Data Analysis For Biologists, 2002. 559 Rathgeber CBK, Santenoise P, Cuny HE. CAVIAR: an R package for checking, displaying and processing 560 wood-formation-monitoring data. Tree Physiol 2018; 38: 1246-1260. 561 Ren P, Rossi S, Camarero JJ, Ellison AM, Liang E, Penuelas J. Critical temperature and precipitation 562 thresholds for the onset of xylogenesis of Juniperus przewalskii in a semi-arid area of the north- 563 eastern Tibetan Plateau. Ann Bot 2018; 121: 617-624. 564 Ren P, Rossi S, Gricar J, Liang E, Cufar K. Is precipitation a trigger for the onset of xylogenesis in Juniperus 565 przewalskii on the north-eastern Tibetan Plateau? Ann Bot 2015; 115: 629-39. 566 Rossi S, Anfodillo T, Menardi R. Trephor: A New Tool for Sampling Microcores from tree stems. IAWA 567 Journal 2006a; 27: 89-97. 568 Rossi S, Deslauriers A, Anfodillo T. Assessment of cambial activity and xylogenesis by microsampling tree 569 species: An example at the alpine timberline. Iawa Journal 2006b; 27: 383-394. 570 Rossi S, Deslauriers A, Anfodillo T, Carraro V. Evidence of threshold temperatures for xylogenesis in 571 conifers at high altitudes. Oecologia 2007; 152: 1-12. 572 Rossi S, Deslauriers A, Griçar J, Seo J-W, Rathgeber CBK, Anfodillo T, et al. Critical temperatures for 573 xylogenesis in conifers of cold climates. Global Ecology and Biogeography 2008; 17: 696-707. 574 Rossi S, Girard MJ, Morin H. Lengthening of the duration of xylogenesis engenders disproportionate 575 increases in xylem production. Glob Chang Biol 2014; 20: 2261-71. 576 Rossi S, Morin H, Deslauriers A. Causes and correlations in cambium phenology: towards an integrated 577 framework of xylogenesis. J Exp Bot 2012; 63: 2117-26. 578 Rotenberg E, Yakir D. Contribution of semi-arid forests to the climate system. Science 2010; 327: 451-4. 579 Saderi S, Rathgeber CBK, Rozenberg P, Fournier M. Phenology of wood formation in larch (Larix decidua 580 Mill.) trees growing along a 1000-m elevation gradient in the French Southern Alps. Annals of 581 Forest Science 2019; 76. 582 Sarris D, Christodoulakis D, KÖRner C. Recent decline in precipitation and tree growth in the eastern 583 Mediterranean. Global Change Biology 2007; 13: 1187-1200. 584 Shamir E, Meko D, Touchan R, Lepley KS, Campbell R, Kaliff RN, et al. Snowpack‐ and Soil Water 585 Content‐Related Hydrologic Indices and Their Association With Radial Growth of Conifers in 586 the Sierra Nevada, California. Journal of Geophysical Research: Biogeosciences 2020; 125. 587 Shen M, Piao S, Cong N, Zhang G, Jassens IA. Precipitation impacts on vegetation spring phenology on 21 588 the Tibetan Plateau. Glob Chang Biol 2015; 21: 3647-56. 589 Stanke H, Finley AO, Domke GM, Weed AS, MacFarlane DW. Over half of western United States' most 590 abundant tree species in decline. Nat Commun 2021; 12: 451. 591 Taynik AV, Barinov VV, Oidupaa OC, Myglan VS, Reinig F, Büntgen U. Growth coherency and climate 592 sensitivity of Larix sibirica at the upper treeline in the Russian Altai-Sayan Mountains. 593 Dendrochronologia 2016; 39: 10-16. 594 Turcotte A, Morin H, Krause C, Deslauriers A, Thibeault-Martel M. The timing of spring rehydration and 595 its relation with the onset of wood formation in black spruce. Agricultural and Forest 596 Meteorology 2009; 149: 1403-1409. 597 Verbancic J, Lunn JE, Stitt M, Persson S. Carbon Supply and the Regulation of Cell Wall Synthesis. Mol 598 Plant 2018; 11: 75-94. 599 Wolf S, Keenan TF, Fisher JB, Baldocchi DD, Desai AR, Richardson AD, et al. Warm spring reduced carbon 600 cycle impact of the 2012 US summer drought. Proc Natl Acad Sci U S A 2016; 113: 5880-5. 601 Wu C, Peng J, Ciais P, Peñuelas J, Wang H, Beguería S, et al. Increased drought effects on the phenology 602 of autumn leaf senescence. Nature Climate Change 2022; 12: 943-949. 603 Wu X, Liu H, Li X, Ciais P, Babst F, Guo W, et al. Differentiating drought legacy effects on vegetation 604 growth over the temperate Northern Hemisphere. Glob Chang Biol 2018; 24: 504-516. 605 Xue BL, Guo Q, Hu T, Xiao J, Yang Y, Wang G, et al. Global patterns of woody residence time and its 606 influence on model simulation of aboveground biomass. Global Biogeochemical Cycles 2017; 607 31: 821-835. 608 Zani D, Crowther TW, Mo LD, Renner SS, Zohner CM. Increased growing-season productivity drives 609 earlier autumn leaf senescence in temperate trees. Science 2020; 370: 1066-+. 610 Zhang D, Yang Y, Lan B. Climate variability in the northern and southern Altai Mountains during the past 611 50 years. Sci Rep 2018a; 8: 3238. 612 Zhang J, Gou X, Alexander MR, Xia J, Wang F, Zhang F, et al. Drought limits wood production of Juniperus 613 przewalskii even as growing seasons lengthens in a cold and arid environment. Catena 2021a; 614 196. 615 Zhang J, Gou X, Pederson N, Zhang F, Niu H, Zhao S, et al. Cambial phenology in Juniperus przewalskii 616 along different altitudinal gradients in a cold and arid region. Tree Physiol 2018b; 38: 840-852. 617 Zhang Y, Xu J, Jiang Y, Mandra T, Rademacher TT, Xue F, et al. Higher plasticity of water uptake in spruce 618 than larch in an alpine habitat of North-Central China. Agricultural and Forest Meteorology 619 2021b; 311. 620 Zhou P, Huang J-G, Liang H, Rossi S, Bergeron Y, Shishov VV, et al. Radial growth of Larix sibirica was 621 more sensitive to climate at low than high altitudes in the Altai Mountains, China. Agricultural 622 and Forest Meteorology 2021; 304-305. 623 Zhou X, Geng X, Yin G, Hänninen H, Hao F, Zhang X, et al. Legacy effect of spring phenology on vegetation 624 growth in temperate China. Agricultural and Forest Meteorology 2020; 281. 625 Ziaco E, Truettner C, Biondi F, Bullock S. Moisture-driven xylogenesis in Pinus ponderosa from a Mojave 626 Desert mountain reveals high phenological plasticity. Plant Cell Environ 2018; 41: 823-836. 627 628 22 629 Supplementary Materials for 630 High variability in the responses of xylem phenology to temperature interacted 631 with precipitation along the arid and semi-arid Southern Altai Mountains 632 633 This file contains： 634 Figures S1-S2 635 Tables S1-S6 636 23 637 Figures: 638 639 Figure S1. Location of the four study sites distributed along the southern Altai 640 Mountain range. AH: Altay high altitude; AL: Altay low altitude; KN: Kanas; QH: 641 Qinghe. 642 643 644 24 645 646 647 Figure S2. Linear mixed model estimates of effects of mean annual temperature (MAT) 648 and total annual precipitation (TAP) on variation in threshold temperatures in L. sibirica 649 (a) and P. obovate(b). 650 651 25 652 Tables: 653 654 Table S1. Study site location and size parameters of sampled tree species. Study Site Site Species Number of Trees DBH Height Code Kanas KN L. sibirica 10 45.43.2 24.61.0 P. obovate 10 46.63.1 24.11.0 Altai low AL L. sibirica 10 27.51.5 12.21.3 altitude P. obovate 10 37.81.7 19.41.2 Altai high AH L. sibirica 5 47.83.0 19.60.7 altitude-H P. obovate 5 53.85.0 20.51.6 Qinghe QH L. sibirica 10 43.52.6 18.10.6 P. obovate 10 41.42.6 18.40.8 655 DBH: diameter at breast height. 656 657 26 658 Table S2. Two-way analysis of variance, with permutation test, in xylogensis threshold 659 temperature in L. sibirica and P. obovate among study sites (S) and between years (Y). Species Parameter df Sum of squares Mean square P L. sibirica Year 1 57.472 57.472 0.000 Site 3 103.255 34.418 0.000 Year×Site 3 3.980 1.327 0.513 Residuals 62 169.050 2.727 P. obovate Year 1 0.327 0.3267 0.765 Site 3 64.126 21.375 0.000 Year×Site 3 30.418 10.140 0.000 Residuals 57 61.563 1.080 660 P-values in bold are <0.05. 661 27 662 Table S3. Linear mixed-effects model estimates of effects of mean annual temperature 663 (MAT) and total annual precipitation (TAP) on onset of xylogenesis in L. sibirica and 664 P. obovate. Species Estimate SE t P L. Intercept -0.211 6.900 -0.031 0.977 sibirica MAT -4.171 1.482 -2.815 0.012 TAP 0.033 0.024 1.330 0.228 MAT × TAP 0.010 0.004 2.423 0.026 P. Intercept 13.778 4.289 3.212 0.017 obovate MAT -4.355 0.845 -5.155 0.000 TAP -0.021 0.015 -1.414 0.184 MAT × TAP 0.013 0.003 5.755 0.000 665 P-values in bold are <0.05. 666 28 667 Table S4. Mean start and end day-of-year dates and duration of xylogenesis events in 668 L. sibirica and P. obovate. Xylogenesis event L. sibirica P. obovate bE 145.8 (119, 166) A 145.0 (131, 165) A bW 164.3 (122, 185) A 161.8 (148, 175) A cE 204.6 (177, 238) A 218.0 (185, 234) B cW 232.3 (218, 256) A 244.1 (219 ,266) B dE 58.8 (18, 98) A 73.1 (43, 93) B dW 68.0 (40, 116) A 82.3 (54, 110) B dX 86.5 (54, 119) A 99.1 (61, 121) B 669 Values in parentheses are minimum and maximum values; different uppercase letters 670 within a row indicate species differences at P < 0.05. bE: start of cell enlargement; bW: 671 start of cell wall thickening; cE: cessation of cell enlargement; cW: cessation of cell 672 wall thickening; dE: duration of cell enlargement; dW: duration of cell wall thickening; 673 and, dX: duration of xylem formation. 674 29 675 Table S5. Generalized linear mixed effects models estimates of effects of mean annual 676 temperature (MAT) and total annual precipitation (TAP) with onset timings of cell 677 enlargement (bE) and cell wall thickening (bW) in L. sibirica and P. obovate. Xylogenesis Species Parameter Estimate SE z P event L. sibirica Intercept 4.861 0.035 139.588 0.000 MAT -0.020 0.006 -3.603 0.000 TAP 0.000 0.000 3.499 0.000 MAT × TAP NS NS NS NS bE P. obovate Intercept 4.979 0.037 136.069 0.000 MAT -0.015 0.006 -2.479 0.013 TAP -0.000 0.000 -0.304 0.761 MAT × TAP NS NS NS NS L. sibirica Intercept 5.008 0.033 152.728 0.000 MAT -0.014 0.005 -2.779 0.005 TAP 0.000 0.000 2.882 0.004 MAT × TAP NS NS NS NS bW P. obovate Intercept 5.083 0.035 146.732 0.000 MAT -0.008 0.006 -1.419 0.156 TAP -0.000 0.000 -0.019 0.985 MAT × TAP NS NS NS NS 678 P-values in bold are <0.05. NS: non-significant interactions at P < 0.05 were removed 679 from the model. 680 30 681 Table S6. Generalized linear mixed effects model estimates of effects of mean annual 682 temperature (MAT) and total annual precipitation (TAP) on cessation timings of cell 683 enlargement (cE) and cell wall thickening (cW) in L. sibirica and P. obovate. Xylogenesis Species Parameter Estimate SE z P event L. sibirica Intercept 5.356 0.031 174.426 0.000 MAT -0.056 0.020 -2.745 0.006 TAP -0.000 0.000 -1.562 0.118 MAT × TAP 0.000 0.000 2.261 0.024 cE P. obovate Intercept 5.369 0.030 179.856 0.000 MAT -0.004 0.005 -0.855 0.392 TAP 0.000 0.000 0.490 0.624 MAT × TAP NS NS NS NS L. sibirica Intercept 5.485 0.028 199.374 0.000 MAT -0.003 0.004 -0.707 0.479 TAP -0.000 0.000 -1.460 0.144 MAT × TAP NS NS NS NS cW P. obovate Intercept 5.470 0.028 193.808 0.000 MAT -0.004 0.005 -0.924 0.356 TAP 0.000 0.000 0.971 0.331 MAT × TAP NS NS NS NS 684 P-values in bold are <0.05. NS: non-significant interactions at P < 0.05 were removed 685 from the model. 686 31