This is the accepted version of the article: Geng, Xiaojun; Fu, Yongshuo H.; Hao, Fanghua; [et al.]. «Climate warming increases spring phenological differences among temperate trees». Global Change Biology, Vol. 26, issue 10 (Oct. 2020), p. 5979-5987. DOI 10.1111/gcb.15301 This version is avaible at https://ddd.uab.cat/record/232632 under the terms of the license 1 Title: Climate warming increases spring phenological differences among temperate 2 trees 3 Running Title: Warming increases spring phenology difference 4 Authors: Xiaojun Geng1, Yongshuo H. Fu1,2*, Fanghua Hao1, Xuancheng Zhou1, Xuan 5 Zhang1, Guodong Yin1, Yann Vitasse3, Shilong Piao4,5,6, Kechang Niu7, Hans J. De 6 Boeck2, Annette Menzel8, Josep Peñuelas9,10 7 Affiliations 1 8 College of Water Sciences, Beijing Normal University, Beijing 100875, China. 2 9 Plants and Ecosystems, Department of Biology, University of Antwerp, Antwerp, 10 Belgium. 3 11 Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), 12 Birmensdorf, Switzerland. 4 13 College of Urban and Environmental Sciences, Sino‐French Institute for Earth 14 System Science, Peking University, Beijing, China. 5 15 Key Laboratory of Alpine Ecology and Biodiversity, Institute of Tibetan Plateau 16 Research, Chinese Academy of Sciences, Beijing, China. 6 17 Center for Excellence in Tibetan Earth Science, Chinese Academy of Sciences, 18 Beijing, China. 7 19 School of Life Science, Nanjing University, Nanjing, China 8 20 Chair of Ecoclimatology, Technische Universität München, D-85354 Freising, 21 Germany 9 22 CREAF, Cerdanyola del Vallès, Barcelona 08193, Catalonia, Spain. 10 23 CSIC, Global Ecology Unit CREAF-CSIC-UAB, Bellaterra, Barcelona 08193, 24 Catalonia, Spain 25 26 Corresponding author: Yongshuo H. Fu 27 Phone: +86-58802736 28 Email: yfu@bnu.edu.cn 29 30 Competing interest statement: The authors declare no competing interests. 31 32 1 33 Abstract 34 Climate warming has substantially advanced spring leaf flushing, but winter chilling 35 and photoperiod co-determine the leaf flushing process in ways that vary among 36 species. As a result, the interspecific differences in spring phenology (IDSP) are 37 expected to change with climate warming, which may in turn induce negative or 38 positive ecological consequences. However, the temporal change of IDSP at large 39 spatio-temporal scales remains unclear. In this study, we analyzed long-term in-situ 40 observations (1951−2016) of six, co-existing temperate tree species from 305 sites 41 across central Europe and found that phenological ranking did not change when 42 comparing the rapidly warming period 1984−2016 to the marginally warming period 43 1951−1983. However, the advance of leaf flushing was significantly larger in early- 44 flushing-species EFS (6.7 ± 0.3 d) than in late-flushing-species LFS (5.9 ± 0.2 d) 45 between the two periods, indicating extended IDSP. This IDSP extension could not be 46 explained by differences in temperature sensitivity between EFS and LFS; however, 47 climatic warming-induced heat accumulation effects on leaf flushing, which were 48 linked to a greater heat requirement and higher photoperiod sensitivity in LFS, drove 49 the shifts in IDSP. Continued climate warming is expected to further extend IDSP 50 across temperate trees, with associated implications for ecosystem function. 51 52 Keywords: climate change, phenological shift, interspecific differences in spring 53 phenology, temperature sensitivity, photoperiod 2 54 Introduction 55 Climate warming impacts terrestrial ecosystem carbon and water cycles (Cannell & 56 Smith, 1986; Chuine, 2010; Bradley, Leopold, Ross & Huffaker, 1999; Snyder & 57 Spano, 2013) over the long term, partly due to earlier spring leaf flushing in temperate 58 and boreal deciduous forests (Basler & Körner, 2014; Chuine, 2010; Fu et al., 2019a; 59 Fu et al., 2019b; Laube et al., 2014). The timing of spring leaf flushing is one of the 60 most sensitive and visible bio-indicators of ongoing climate change (Kharouba et al., 61 2018; Van Asch & Visser, 2007; Zohner, Benito, Svenning & Renner, 2016) because 62 it is principally controlled by temperature (Peñuelas, Rutishauser & Filella, 2009; 63 Zohner, Mo & Renner, 2018). However, other factors, such as winter low temperature 64 (or chilling accumulation) (Keenan et al., 2014; Piao et al., 2019), and photoperiod 65 (day length) (Chuine, 2010; Körner & Basler, 2010), also affect the timing of leaf 66 flushing and their importance may increase under further warming (Menzel et al., 67 2006). Species differences in forcing, chilling and photoperiodic requirements (Chuine 68 et al., 2016; Fu et al., 2014; Keenan et al., 2014) may result in altered interspecific time 69 differences in leaf flushing phenology (IDSP), which may ultimately alter the sequence 70 in time of leaf flushing (“ranking”, previously also known as interception (Peñuelas et 71 al., 2009)) among the different species. There are three theoretical shifts in IDSP 72 between early and late flushing species (Vitasse, Lenz, Hoch & Körner,2014) (called 73 hereafter EFS and LFS, respectively, ranking based on the timing of leaf-out) from 74 current to future climatic conditions (Fig. S1a): stable IDSP, where leaf flushing 75 advance is similar between EFS and LFS (Fig. S1b); reduced IDSP between EFS and 3 76 LFS, due to greater leaf flushing advance for LFS than EFS (Fig. S1c); and extended 77 IDSP between EFS and LFS, due to greater leaf flushing advance for EFS than LFS 78 (Fig. S1d). 79 80 Changes in IDSP may elicit ecological perturbations. For example, plant competition 81 may change, as species with greater advance in leaf flushing may profit from a longer 82 growing season, shade competitors with later flushing and thus gain a competitive 83 advantage providing that they can still avoid damaging late spring frost events. 84 Changes in competitive relationships or trophic mismatches could further influence 85 resource allocation and structural adjustment of the ecosystem, eventually altering 86 terrestrial carbon and water cycling. For example, leaf phenology directly affects the 87 light and water requirements in the forest canopy and understory, which further 88 determines the photosynthetic efficiency and carbon uptake and metabolism (Cleland, 89 Chuine, Menzel, Mooney & Schwartz, 2007; Peñuelas, Rutisshauser & Filella, 2009; 90 Schwartz, 1998). Reductions in spatial variation of leaf flushing have been reported 91 across central Europe within species both across elevations (Vitasse, Signarbieux & Fu, 92 2018) and latitudes (Ma, Huang, Hänninen & Berninger, 2018), which was also 93 confirmed by remote sensing analyses associated with climatic warming across the 94 Northern hemisphere (Liu et al., 2019). However, temporal changes in IDSP among 95 species have been little studied (but see Peñuelas & Filella, 2001; Roberts, Tansey, 96 Smithers & Phillimore, 2015) despite their importance in understanding ecosystem 97 responses to ongoing climate change. 4 98 Therefore, we investigated the IDSP among six common, co-existing tree species, three 99 of them EFS and three LFS, across 305 central European sites with temperate forest 100 tree species (Fig. S2). The aim of our study was to determine and quantify temporal 101 changes and associated physiological and climatic drivers in leaf flushing IDSP among 102 the six species during the period 1951−2016. 103 104 Material and methods 105 Datasets. Leaf flushing data for EFS Betula pendula (BP, European silver birch), 106 Aesculus hippocastanum (AH, horse chestnut), and Alnus glutinosa (AG, European 107 alder), and LFS Fagus sylvatica (FS, European beech), Quercus robur (QR, Pedunculate 108 oak), and Fraxinus excelsior (FE, European ash) co-occurring in central European sites 109 during the period 1951–2016 were obtained from the open access Pan European 110 Phenology network (www.pep725.eu). It is one specimen per site for each species, and 111 a specimen presents a group of individuals (2-3) at each site, and then the averaged dates 112 were defined at the phenological dates (DWD, 2018). Datasets comprised a minimum 113 of 30 years over the period 1951-2016, representing 15 years in each of two 33-year 114 periods (mild warming period of 1951−1983 and pronounced warming period of 115 1984−2016, Fig. S3), for all six species at each site. Data >2×SD (standard deviation of 116 the mean leaf flushing date) were excluded to discard possible errors, which account for 117 3.8% in EFS and 4.0% in LFS during 1951-2016, distributed randomly. Similar results 118 without data reduction were found (Fig. S4). In total, we selected data for 305 sites (Fig. 119 S2), to which a gridded climatic dataset was applied, with a spatial resolution of 0.25° 5 120 (approximately 25 km, Beer et al., 2014), comprising daily mean air temperature, 121 precipitation, and radiation measurements (Fu et al., 2015). We further tested the 122 reliability of the results using the well-known Princeton climatic dataset, at 0.25° 123 spatial resolution, (http://hydrology.princeton.edu/data/pgf/v3/0.25deg /daily/), and 124 very similar results were found (Fig. S5). 125 126 Analysis. We analyzed spring warming as the mean value and the corresponding SD 127 of the air temperature during the temperature relevant period (TRP). TRP was 128 calculated as the period before leaf flushing with the highest absolute partial correlation 129 between leaf flushing date and mean air temperature. Considering the differences 130 among species and sites in responding to climate change, different TRPs were used for 131 each species and site (Fig. S6), like in earlier analyses (Fu et al., 2015), ranging from 132 15 to 120 d, with 15-d steps. We investigated the TRP variation with latitudes (Fig. S6). 133 The mean length of TRP was 59 days prior to the day of leaf flushing, but a large 134 variation was found among sites and species, with TRP decreasing with increasing 135 latitudes (Fig. S6), thus showing the robustness of the location-specific application of 136 the TRP in the present study. 137 138 Apparent temperature sensitivity (ST) was defined as the number of advanced days per 139 1 ‐ of warming, which has been widely used to evaluate the long-term implications of 140 temperature changes for plant phenology. The ST was determined by reduced major- 141 axis regression (RMA) between leaf flushing date and mean air temperature during the 6 142 TRP (Keenan, Richardson & Hufkens, 2019). To validate the results, we also used daily 143 maximum temperatures to analyze TRP and apparent sensitivities of leaf flushing to 144 maximum temperature during TRP (STmax). Similar results were found (Fig. S7). 145 146 Chilling accumulation breaks dormancy in deciduous tree species and heat 147 accumulation leads to leaf flushing, so their metrics have been widely used to estimate 148 the physiological requirement of spring leaf flushing in temperate woody species. 149 Chilling requirement (CDreq ) was defined as the number of days with non-freezing 150 mean air temperature (generally between 0 and 5 °C) during the period 1st November 151 to the mean leaf flushing date: 𝐿𝑂 152 CDreq (t)= ∑ 1 if 0 ≤ Tt ≤ 𝑇ℎ𝑖𝑔ℎ t0 153 where t0 is the start date of chilling accumulation (fixed at 1st November before the year 154 of leaf flushing), LO is the mean leaf flushing date, Tt is mean daily air temperature, 155 and Thigh is the upper limit of the temperature threshold (generally 5 or 8 °C). We tested 156 the same calculation by setting the temperature threshold within 0 to 8 °C and tested 157 1st September as start date (Fig. S8); In addition, because the correlation between 158 chilling and forcing is still unclear, to test the robustness of the chilling estimation, we 159 also calculated the chilling accumulation by fixing the time period from 1st November 160 to 31th December (Fig. S8d). This led to very similar results, so we reported the results 161 using the threshold of 5 °C and 1st November until the start date of leaf flushing. 162 7 163 The heat requirement for leaf flushing was defined as the growing degree days (GDD), 164 calculated as the sum of mean air temperature above a temperature threshold from 1st 165 January to leaf flushing date. We used three common methods as follows: 166 (1) Linear function, by using a base temperature of 5 °C: 𝐿𝑂 167 𝐺𝐷𝐷 = ∑(𝑇𝑡 − 5) 𝑖𝑓 𝑇𝑡 ≥ 5 𝑡0 168 (2) Piecewise function, by using a temperature threshold of 5 and 15 °C, and fixed 169 units when temperature was above 15 °C: 𝐿𝑂 0 𝑖𝑓 𝑇𝑡 < 5 170 𝐺𝐷𝐷 = ∑ { 𝑇𝑡 − 5 𝑖𝑓 5 ≤ 𝑇𝑡 ≤ 15 𝑡0 10 𝑖𝑓 Tt > 15 171 (3) Sigmoidal function, by using a base temperature of 0 °C: 𝐿𝑂 0 𝑖𝑓 𝑇𝑡 < 0 172 𝐺𝐷𝐷 = ∑ { 28.4 𝑖𝑓 𝑇𝑡 ≥ 0 𝑡0 1 + 𝑒 −0.185(𝑇𝑡−18.4) 173 where t0 is the start date of GDD calculation, fixed at 1st January, LO is the leaf flushing 174 date, and Tt is daily mean air temperature. The three methods produced similar results, 175 so we only reported the results of the linear function in the main text and reported the 176 results of the other two functions in the appendix (Fig. S9). 177 178 Statistical analysis. We then determined the frequency distributions of main variables 179 (difference in leaf flushing dates / temperature sensitivity / chilling accumulation / 8 180 GDD requirements between EFS and LFS or between 1951-1983 and 1984-2016), and 181 plotted in histograms. For each species, we applied paired t-tests to detect whether the 182 average flushing date and its variance changed significantly between time periods. To 183 adequately model the structure in the data (e.g. site locations, species, years), we 184 adopted the mixed effect models using lmerTest package in R3.5.2 (Phillimore, Leech, 185 Pearce-Higgins & Hadfield, 2016), by setting leaf flushing dates, temperature 186 sensitivity, chilling accumulation and GDD requirements as response variables, setting 187 time period (1951-1983 versus 1984-2016) and grouping of species (EFS versus LFS) 188 as fixed effects, and setting sites (including latitude, longitude and elevation), species 189 and years as random effects (Table S1). We also analyzed the number of sites in which 190 species phenological shift occurred and the direction of change, and presented the 191 results through CIRCOS figures (Fig. 1c) by http://mkweb.bcgsc.ca/tableviewer/. 192 193 Results and discussion 194 The mean leaf flushing date over the period 1951−2016 was 21 April (Day of year: 111 195 ± 6, mean ± SD across sites and species) for the EFS Betula pendula (BP), Aesculus 196 hippocastanum (AH), and Alnus glutinosa, (AG), and 4 May (Day of year: 124 ± 5 d) 197 for the LFS Fagus sylvatica (FS), Quercus robur (QR), and Fraxinus excelsior (FE) 198 (Fig. 1a). Thus, there was a difference of 13 d in mean leaf flushing dates between EFS 199 and LFS. Variation in mean leaf flushing date (measured as the standard deviation of 200 leaf flushing date over the study period, 1×SD) tended to be greater for EFS (9.4 ± 2.5 201 d) than for LFS (7.8 ± 2.0 d) (P < 0.05, Fig. 1a). We found that the advance in leaf 9 202 flushing date across the six temperate tree species increased from 0.9 ± 0.7 d (mean ± 203 s.e.) during the period 1951−1983 to 13.5 ± 0.7 d during the period 1984−2016 , when 204 climate warming accelerated (the species-specific temporal changes in leaf flushing 205 dates across all sites between the two time period can be found in Table S2 and Fig. 206 S10), supporting previous reports (Fu et al., 2019b; Menzel, 2013). Although the 207 magnitude of advance differed among the six species, the overall phenological ranking 208 remained stable over the period 1951-2016 (Fig. S10) while site-specific shifts in 209 ranking occurred at 49.2-52.8% of sites for EFS and 16.1-33.1% of sites for LFS (Fig. 210 1c). When compared the leaf flushing dates between 1951−1983 and 1984−2016, we 211 found a greater magnitude of advancement in the EFS (mean ± s.e.: 6.7 ± 0.3 d) than 212 in the LFS (5.9 ± 0.2 d) (Fig. 1b, mixed effect model was applied including sites and 213 species as random effects: P < 0.001), confirming an enlarged IDSP between 214 1951−1983 and 1984−2016. 215 216 To explain our finding of extended IDSP, but general stability in rank order of leaf 217 flushing among the six species, we propose four mutually non-exclusive hypotheses: 218 (1) difference in seasonal warming, since warming may be stronger during earlier 219 spring than in later spring; (2) difference in temperature sensitivity, where temperature 220 sensitivity to leaf flushing may be greater in EFS than LFS (Menzel, Estrella & Fabian, 221 2001; Menzel, Sparks, Estrella & Roy, 2006); (3) interaction between chilling effects 222 and heat requirements among species, where chilling effects are similar due to identical 223 winter conditions, but LFS may have greater heat requirement than EFS (Fu et al., 10 224 2015); and (4) differences in photoperiod and water limitation among species, where 225 sensitivity to photoperiod and water shortage may be greater in LFS than EFS (Chuine, 226 2010; Cleland et al., 2007), leading to a smaller advance in LFS under climate warming. 227 228 To test the first hypothesis, we analyzed monthly warming trends over the study period, 229 but did not find early spring (February and March) significantly more strongly warming 230 (P > 0.05) than later spring (April and May); in contrast, climate warming in May was 231 greater than in February and March (Fig. 2a). Next, we estimated mean temperature 232 and its variability (1×SD) during the most temperature relevant period (TRP, see 233 Methods) for leaf flushing during the periods 1984−2016 and 1951−1983, and found 234 increased temperatures during the TRP for both groups of tree species (EFS and LFS), 235 but in contrast to the hypothesis, because the increase was greater in LFS (mean 236 increase: 0.9 °C) than in EFS (mean increase: 0.7°C) (Fig. 2b). Variation in temperature 237 during the most relevant period remained stable for EFS, but increased for LFS (Fig. 238 2c). A higher fluctuated temperature may constrain advances of leaf flushing of LFS to 239 avoid frost damages (Zohner et al., 2017). These results indicate that the climate 240 variation, more than warming, may drive the extended time difference in leaf flushing 241 between EFS and LFS, assuming that the temperature sensitivity of EFS and LFS was 242 identical. 243 244 It is more likely that sensitivity to temperature changes is greater in EFS than in LFS 245 in order to enable competitive advantages, although earlier leaf flushing may also 11 246 increase the risk of late frost damage (Inouye, 2008; Richardson et al., 2018). Therefore, 247 a greater advancement in leaf flushing may be expected in EFS, given stable warming 248 trends. Indeed, we found significantly larger ST (apparent temperature sensitivity) of 249 leaf flushing in EFS than in LFS (3.8 ± 0.1 and 2.9 ± 0.1 d oC-1, respectively) during 250 the period 1951−1983 (P < 0.05, mixed effect model was applied including sites and 251 species as random effects); however, ST was significantly reduced to 2.9 ± 0.1 d °C-1 in 252 EFS during the period 1984−2016, but remained stable in LFS. As a result, there was 253 no overall difference in ST between EFS and LFS during the recent period 1984−2016 254 (Fig. 3d, mixed effect model: P = 0.14). Reduced ST in EFS and a trend towards higher 255 temperatures at leaf flushing in LFS rather than in EFS (Fig. 2b) did not explain the 256 extended time difference in leaf flushing between the two groups, because a warmer 257 late spring and similar ST for EFS and LFS during the period 1984−2016 should have 258 led to the greater advancement in leaf flushing for LFS. 259 260 The third hypothesis relates to physiological processes associated with chilling and 261 subsequent heat requirements for spring leaf flushing. Climate warming may reduce 262 the plants’ ability to fulfill the amount of chilling that is required to break dormancy in 263 many temperate tree species (Hänninen, 2016; Piao et al., 2019), with the consequence 264 that greater heat is required to trigger leaf flushing (Cannell & Smith,1983; Fu et al., 265 2015). EFS and LFS experience the same length of chilling days during the winter, but 266 EFS and LFS experience different amounts of chilling accumulation during the spring. 267 Our data support this hypothesis, because the difference in chill days between EFS and 12 268 LFS during the period 1951−1983 (2.0 ± 0.1 d) was not significantly different from the 269 period 1984−2016 (1.8 ± 0.1 d) (Fig. 3a, P = 0.62, mixed effect model was applied 270 including sites and species as random effects), but species’ heat requirements, defined 271 as growing degree days (GDD, see Methods), were negatively related to chilling 272 accumulation and significantly increased in both EFS (1951−1983: 113.1 ± 1.9 °C; 273 1984−2016: 118.2 ± 2.0 °C) and LFS (1951−1983: 173.2 ± 2.5 °C; 1984−2016: 183.4 274 ± 2.7 °C) (Fig. 3b). This larger increase in GDD in LFS led to increased differences in 275 GDD requirements between the two groups (1951−1983: 60.1 ± 1.0 °C; 1984−2016: 276 65.3 ± 1.1 °C) (Fig. 3b, P < 0.01). The time required to accumulate an increased GDD 277 in LFS may subsequently result in a smaller advance in leaf flushing in LFS and extend 278 the time difference in leaf flushing between the two groups. Thus, the extended IDSP 279 is likely to occur due to the asymmetric increase in heat requirements between EFS and 280 LFS. 281 282 Our fourth hypothesis relates to effects of photoperiod on the leaf flushing process that 283 may vary with species due to genotype and ontogeny (Chuine et al., 2016; Flynn & 284 Wolkovich, 2018; Richardson, Hufkens, Li & Ault, 2019; Yang & Rudolf, 2010); for 285 example, it has been suggested that LFS rely on photoperiod to a greater extent than 286 EFS (Van Asch & Visser, 2007). We found that climate warming advanced leaf 287 flushing, so the effects of a resulting shorter photoperiod on the leaf flushing process 288 may have increased in both EFS and LFS, but larger in LFS, and thus led to a relatively 289 shorter advancement in leaf flushing date. Since it was not possible to assess direct 13 290 effects of photoperiod in this study, we estimated changes in variation of leaf flushing 291 dates for both EFS and LFS between the two study periods as a surrogate measure, 292 because a higher photoperiod effect is likely to be associated with a lower variation in 293 flushing dates (the variation of flushing dates was defined as one standard deviation: 294 1×SD, LOSD). We found that the LOSD tended to be lower during the period 1984−2016 295 (EFS: 7.4 ± 1.8 d; LFS: 5.9 ± 1.6 d) than during 1951−1983 (EFS: 8.1 ± 1.7 d; LFS: 296 6.1 ± 1.5 d) (Fig. 3c) for both EFS and LFS, but the reduction was significantly lower 297 in LFS (0.2 ± 0.1 d) than in EFS (0.7 ± 0.1 d), suggesting a larger photoperiod effect 298 in LFS. To further test the photoperiod limitation hypothesis, we investigated the 299 correlation between latitudes and IDSP comparing EFS and LFS, as well as the species- 300 specific variation of advanced leaf flushing date along latitude. Interestingly, we found 301 the IDSP was significantly reduced at lower latitudes across all species (Fig. 4), and 302 the advancement increased at higher latitudes for each species (P < 0.05, Fig. S11). The 303 significant reduction in advancement at low latitudes suggested that photoperiod likely 304 limits the advancement at the low latitudes, rather than at high latitudes, to constrain 305 too early leaf flushing. Furthermore, a relatively smaller advancement in leaf flushing 306 was found at higher latitudes for LFS than EFS, which may be because LFS is more 307 sensitive to photoperiod than EFS at higher latitudes. These results suggest that the 308 extended IDSP difference between EFS and LFS may partially explained by the 309 photoperiod. In addition, within forests, earlier flushing of EFS may increase the water 310 stress for LFS (Bradley et al., 1999; Spano, Snyder & Cesaraccio, 2013), and thus 311 postpone the date of leaf flushing of LFS (Spano, Snyder & Cesaraccio, 2013). 14 312 However, annual precipitation over the study area is larger than 700 mm (EEA, 2012), 313 and the difference in precipitation sum over the TRP between EFS and LFS was 314 insignificant (Fig. S12). This suggests that water shortage limitations on spring leaf- 315 out for LFS were unlikely, rendering a contribution to the extended IDSP between EFS 316 and LFS improbable. 317 318 Phenological sensitivity to environmental changes is a strategy of temperate trees to 319 optimize growth in regards to climatic fluctuations. Our results provide empirical 320 evidence for an overall stability in phenological ranking among temperate tree species 321 in central Europe over the past six decades. Interestingly, we found that differences in 322 phenological responses between EFS and LFS under rapid warming resulted in a 323 significantly extended IDSP. Plant phenology plays a key role in ecosystem structure 324 and function, and changes in IDSP are likely to elicit ecological consequences (Cleland 325 et al., 2007; Van Asch & Visser, 2007; Vitasse et al., 2018). The advancement in leaf 326 flushing in EFS may be exacerbated under future climate warming conditions to a 327 greater extent than in LFS, so that EFS may benefit from a longer growing season and 328 gain a competitive advantage over LFS in light and water accessibility. Shifts in IDSP 329 may also affect trophic interactions, because of the reliance of many vertebrates on 330 leaf-feeding insects for food. However, great variation exists in the speed of 331 phenological adjustments of insects, possibly leading to trophic synchrony disruption 332 (Gaudry et al., 2015; Renner & Zohner., 2018). Further studies are warranted to 333 confirm these results for different tree species and in different climate zones, especially 15 334 in regions with larger climatic ranges, such as North America and Alpine ecosystems. 335 336 Under future climate warming conditions, differences in timing of spring leaf flushing 337 among species may continue to increase, and may result in a change in the order of 338 spring phenology in temperate forests, especially due to species-specific chilling 339 requirement, as argued by Roberts et al. (2015) using modeling approaches. 340 Furthermore, as extreme weather events become more frequent, it is also worth noting 341 that chilling and GDD requirements for some tree species would also be altered, which 342 will further complicate the model development. However, accurate prediction of the 343 development of phenological difference under future climate change remains a 344 challenge and requires additional investigation, because chilling and GDD 345 requirements, and the temperature thresholds in chilling and GDD estimation, differed 346 at species- and site-scales, and their correlations with photoperiod effects are still 347 uncertain. Ecosystem-scale impacts of enlarged IDSP, for example, effects of extended 348 phenological differences on the carbon and water balance and on nutrient losses, are 349 largely unknown. In LFS-dominated forests, like in our study area (EEA, 2017), the 350 extended IDSP would likely hamper increases in the forests’ carbon sink strength 351 because the LFS would not be able to extend their leaf season to the same extent than 352 EFS. In addition, the extended IDSP may increase uncertainty in spring phenology 353 estimation from satellite images, as remote sensing approaches mainly trace the 354 phenology of EFS (Fu et al., 2019a), and are subsequently prone to overestimation of 355 ecosystem carbon uptake (Piao, Friedlingstein, Ciais, Viovy & Demarty, 2007). In 16 356 addition, previous studies have estimated that earlier start of growing season will 357 increase the vegetation productivity (Zohner, Mo, Pugh, Bastin & Crowther, 2020). 358 However, when considering the spatial variability of phenology shifts and the possible 359 mutual constrains caused by interspecific differences in leaf flushing, the prediction of 360 ecosystem productivity becomes more uncertain. Therefore, we encourage further 361 research that focuses on potential implications of shifted IDSP for ecosystem carbon 362 and water balance, as well as for plant−animal interactions, where experimental studies 363 would be particularly insightful to improve our understanding of shifted IDSP 364 consequences for ecosystems. 17 365 Acknowledgements 366 This study was supported by the General program of National Nature Science 367 Foundation of China (No. 31770516), the National Key Research and Development 368 Program of China (2017YFA06036001), and the 111 Project (B18006) and 369 Fundamental Research Funds for the Central Universities (2018EYT05). Josep 370 Peñuelas acknowledges support from the European Research Council through Synergy 371 grant ERC-2013-SyG-610028 “IMBALANCE-P”. Annette Menzel acknowledges 372 support by the Bavarian Ministry of Science via the Bavarian Climate Research 373 Network (bayklif / BAYSICS). The author thanks all members of the PEP725 project 374 for providing the phenological data. The authors thank Dr. Susanne S. Renner, Dr. Ivan 375 Janssens, Dr. Quanguo Zhang and Dr. Ally Phillimore for their constructive feedback 376 during the interactive discussion that helped to improve the manuscript. Y.H.F. 377 designed the research and drafted the paper, X.J.G performed the analyses and all 378 authors contributed to the interpretation of the results and to the text. 18 379 References 380 Basler, D., & Körner, C. (2014). Photoperiod and temperature responses of bud swelling and bud burst 381 in four temperate forest tree species. Tree Physiology, 34(4), 377-388. 382 https://doi.org/10.1093/treephys/tpu021 383 Beer, C., Weber, U., Tomelleri, E., Carvalhais, N., Mahecha, M. & Reichstein, M. (2014). 384 Harmonized European long-term climate data for assessing the effect of changing 385 temporal variability on land-atmosphere CO2 fluxes. Journal of Climate, 27, 4815–4834. 386 http://dx.doi.org/10.1175/JCLI-D-13-00543.1 387 Bradley, N.L., Leopold, A. C., Ross, J., & Huffaker, W. (1999). Phenological changes reflect climate 388 change in Wisconsin. Proceedings of the National Academy of Sciences, 96, 9701-9704. 389 https://doi.org/10.1073/pnas.96.17.9701 390 Cannell, M. G. R., & Smith, R. I. (1983). Thermal time, chill days and prediction of budburst in Picea 391 sitchensis. Journal of Applied Ecology, 3(20), 951-963. https://doi.org/10.2307/2403139 392 Cannell, M., & Smith, R. I. (1986). Climate warming, spring budburst and frost damage on trees. Journal 393 of Applied Ecology, 23(1), 177-191. https://doi.org/ 10.2307/2403090 394 Chuine, I. (2010). Why does phenology drive species distribution? Philosophical Transactions of the 395 Royal Society B: Biological Sciences, 365(1555), 3149-3160. 396 https://doi.org/10.1098/rstb.2010.0142 397 Chuine, I., Bonhomme, M., Legave, J., de Cortázar‐ Atauri, I. G., Charrier, G., Lacointe, A., & Ameglio, 398 T. (2016). Can phenological models predict tree phenology accurately in the future? The 399 unrevealed hurdle of endodormancy break. Global Change Biology, 22(10), 3444-3460. 400 https://doi.org/10.1111/gcb.13383 401 Cleland, E. E., Chuine, I., Menzel, A., Mooney, H., & Schwartz, M. (2007). Shifting plant phenology in 402 response to global change. Trends in Ecology & Evolution, 22(7), 357-365. 403 https://doi.org/10.1016/j.tree.2007.04.003 404 DWD (2018) Vorschriften und Betriebsunterlagen für die phänologischen Beobachter des Deutschen 405 Wetterdienstes VuB 17. Offenbach am Main. 406 European Environment Agency report: climate change, impacts and vulnerability in Europe. (2012). 407 International Journal of Climate Change Strategies and Management, 5(2). 408 https://doi.org/10.1108/ijccsm.2013.41405baa.003 409 European Environment Agency report: forest utilization. (2017). 410 (https://www.eea.europa.eu/airs/2017/natural-capital/forest-utilisation). 411 Flynn, D. F. B., & Wolkovich, E. M. (2018). Temperature and photoperiod drive spring phenology 412 across all species in a temperate forest community. New Phytologist, 219(4), 1353-1362. 413 https://doi.org/10.1111/nph.15232 414 Fu, Y. H., Piao, S., Op De Beeck, M., Cong, N., Zhao, H., Zhang, Y., ... Janssens, I. A. (2014). Recent 415 spring phenology shifts in western Central Europe based on multiscale observations. Global 416 Ecology and Biogeography, 23(11), 1255-1263. https://doi.org/10.1111/geb.12210 417 Fu, Y. H., Piao, S., Vitasse, Y., Zhao, H., De Boeck, H. J., Liu, Q., ... Janssens, I. A. (2015). Increased 418 heat requirement for leaf flushing in temperate woody species over 1980-2012: effects of chilling, 419 precipitation and insolation. Global Change Biology, 21(7), 2687-2697. 420 https://doi.org/10.1111/gcb.12863 421 Fu, Y. H., Piao, S., Zhou, X., Geng, X., Hao, F., Vitasse, Y., ... Janssens, I. A. (2019a). Short photoperiod 19 422 reduces the temperature sensitivity of leaf-out in saplings of Fagus sylvatica but not in horse 423 chestnut. Global Change Biology, 25(5), 1696-1703. https://doi.org/10.1111/gcb.14599 424 Fu, Y. H., Zhang, X., Piao, S., Hao, F., Geng, X., Vitasse, Y., ... Janssens, I. A. (2019b). Daylength helps 425 temperate deciduous trees to leaf-out at the optimal time. Global Change Biology. 25, 2410–2418. 426 https://doi.org/10.1111/gcb.14633 427 Fu, Y. H., Zhao, H., Piao, S., Peaucelle, M., Peng, S., Zhou, G., … Janssens, I. A. (2015). Declining 428 global warming effects on the phenology of spring leaf unfolding. Nature, 526, 104-107. 429 https://doi.org/10.1038/nature15402 430 Gaudry, W., Saïd, S., Gaillard, J. M., Chevrier, T., Loison, A., Maillard, D., & Bonenfant, C. (2015). 431 Partial migration or just habitat selection? Seasonal movements of roe deer in an Alpine population. 432 Journal of Mammalogy, (96), 502-510. https://doi.org/10.1093/jmammal/gyv055 433 Hänninen, H. (2016). Boreal and Temperate Trees in a Changing Climate. Dordrecht, Netherlands: 434 Springer Business Media. 435 Heberling, J. M., MacKenzie, C. M., Fridley, J. D., Kalisz, S., & Primack, R. B. (2019). Phenological 436 mismatch with trees reduces wildflower carbon budgets. Ecology Letters, 22(4), 616-623. 437 https://doi.org/10.1111/ele.13224 438 Inouye, D. W. (2008). Effects of Climate change on phenology, frost damage, and floral abundance of 439 montane wildflowers. Ecology, 89(2), 353-362. https://doi.org/10.1890/06-2128.1 440 Keenan, T. F., Gray, J., Friedl, M. A., Toomey, M., Bohrer, G., Hollinger, D. Y., ... Richardson, A. D. 441 (2014). Net carbon uptake has increased through warming-induced changes in temperate forest 442 phenology. Nature Climate Change, 4(7), 598-604. https://doi.org/10.1038/nclimate2253 443 Keenan, T. F., Richardson A. D. & Hufkens, K. (2019). On quantifying the apparent temperature 444 sensitivity of plant phenology. New Phytologist, 225 (2): 1033-1040. 445 https://doi.org/10.1111/nph.16114 446 Kharouba, H. M., Ehrlén, J., Gelman, A., Bolmgren, K., Allen, J. M., Travers, S. E., ... Wolkovich, E. 447 M. (2018). Global shifts in the phenological synchrony of species interactions over recent decades. 448 Proceedings of the National Academy of Sciences of the United States of America, 115(20), 5211- 449 5216. https://doi.org/10.1073/pnas.1714511115 450 Körner, C., & Basler, D. (2010). Phenology under global warming. Science, 327, 1461–1462. 451 https://doi.org/10.1126/science.1186473 452 Laube, J., Sparks, T. H., Estrella, N., Höfler, J., Ankerst, D. P., & Menzel, A. (2014). Chilling outweighs 453 photoperiod in preventing precocious spring development. Global Change Biology, 20(1), 170- 454 182. https://doi.org/10.1111/gcb.12360 455 Liu, Q., Piao, S., Fu, Y. H., Gao, M., Peñuelas, J., ... Janssens, I. A. (2019). Climatic warming increases 456 spatial synchrony in spring vegetation phenology across the northern hemisphere. Geophysical 457 Research Letters, 46(3), 1641-1650. https://doi.org/10.1029/2018GL081370 458 Ma, Q., Huang, J., Hänninen, H., & Berninger, F. (2018). Reduced geographical variability in spring 459 phenology of temperate trees with recent warming. Agricultural and Forest Meteorology, 256, 460 526-533. https://doi.org/10.1016/j.agrformet.2018.04.012 461 Menzel, A. (2013) Europe. In: Schwartz, M. D. (Eds.), Phenology: an integrative environmental science 462 (pp. 53-65). Dordrecht Heidelberg New York London: Springer. 463 Menzel, A., Estrella, N., & Fabian, P. (2001). Spatial and temporal variability of the phenological 464 seasons in Germany from 1951 to 1996. Global Change Biology, 7, 657-666. 465 https://doi.org/10.1111/j.1365-2486.2001.00430.x 20 466 Menzel, A., Sparks, T. H., Estrella, N., Koch, E., Aasa, A., Ahas, R., ... Zust, A. (2006). European 467 phenological response to climate change matches the warming pattern. Global Change Biology, 468 12(10), 1969-1976. https://doi.org/10.1111/j.1365-2486.2006.01193.x 469 Menzel, A., Sparks, T. H., Estrella, N., & Roy, D. B. (2006). Altered geographic and temporal variability 470 in phenology in response to climate change. Global Ecology and Biogeography, 15, 498-504. 471 https://doi.org/10.1111/j.1466-822x.2006.00247.x 472 Peñuelas, J., Rutishauser, T., & Filella, I. (2009). Phenology feedbacks on climate change. Science, 473 324(5929), 887-888. https://doi.org/10.1126/science.1173004 474 Peñuelas, J., & Filella, I. (2001). Responses to a warming world. Science, 294(5543), 793. 475 https://doi.org/10.1126/science.1066860 476 Phillimore, A. B., Leech, D. I., Pearce-Higgins, J. W., & Hadfield, J. D. (2016). Passerines may be 477 sufficiently plastic to track temperature-mediated shifts in optimum lay date. Global Change 478 Biology, 22(10), 3259-3272. https://doi.org/10.1111/gcb.13302 479 Piao, S., Friedlingstein, P., Ciais, P., Viovy, N., & Demarty, J. (2007). Growing season extension and 480 its impact on terrestrial carbon cycle in the Northern Hemisphere over the past 2 decades. Global 481 Biogeochemical Cycles, 21(3), GB3018. https://doi.org/10.1029/2006GB002888 482 Piao, S., Liu, Q., Chen, A., Janssens, I. A., Fu, Y., Dai, J., ... Zhu, X. (2019). Plant phenology and global 483 climate change: Current progresses and challenges. Global Change Biology, 25(6), 1922-1940. 484 https://doi.org/10.1111/gcb.14619 485 Renner, S. S. & Zohner, C. M. (2018). Climate change and phenological mismatch in trophic interactions 486 among plants, insects, and vertebrates. Annual Review of Ecology, Evolution, and Systematics, 49 487 (1): 165-182. https://doi.org/10.1146/annurev-ecolsys-110617-062535 488 Richardson, A. D., Hufkens, K., Li, X., & Ault, T. R. (2019). Testing Hopkins’ bioclimatic law with 489 PhenoCam data. Applications in Plant Sciences, 7(3), e1228. https://doi.org/10.1002/aps3.1228 490 Richardson, A. D., Hufkens, K., Milliman, T., Aubrecht, D. M., Furze, M. E., Seyednasrollah, B., ... 491 Hanson, P. J. (2018). Ecosystem warming extends vegetation activity but heightens vulnerability 492 to cold temperatures. Nature, 560(7718), 368. https://doi.org/10.1038/s41586-018-0399-1 493 Roberts, A. M. I., Tansey, C., Smithers, R. J., & Phillimore, A. B. (2015). Predicting a change in the 494 order of spring phenology in temperate forests. Global Change Biology, 21(7), 2603-2611. 495 https://doi.org/10.1111/gcb.12896 496 Schwartz, M. D. (1998). Green-wave phenology. Nature, 394(6696), 839-840. 497 https://doi.org/10.1038/29670 498 Snyder, R. L. & Spano, D. (2013) Phenology and Evapotranspiration. In: Schwartz, M. D. (Eds.), 499 Phenology: an integrative environmental science (pp. 521-538). Dordrecht Heidelberg New York 500 London: Springer. 501 Spano, D., Snyder, R. L. & Cesaraccio, C. (2013) Mediterranean Phenology. In: Schwartz, M. D. (Eds.), 502 Phenology: an integrative environmental science (pp. 173-196). Dordrecht Heidelberg New York 503 London: Springer. 504 Van Asch, M., & Visser, M. E. (2007). Phenology of forest caterpillars and their host trees: The 505 importance of synchrony. Annual Review of Entomology, 52(1), 37-55. 506 https://doi.org/10.1146/annurev.ento.52.110405.091418 507 Vitasse, Y., Lenz, A., Hoch, G., & Körner, C. (2014). Earlier leaf-out rather than difference in freezing 508 resistance puts juvenile trees at greater risk of damage than adult trees. Journal of Ecology, 102(4), 509 981-988. https://doi.org/10.1111/1365-2745.12251 21 510 Vitasse, Y., Signarbieux, C., & Fu, Y. H. (2018). Global warming leads to more uniform spring 511 phenology across elevations. Proceedings of the National Academy of Sciences of the United States 512 of America, 115(5), 1004-1008. https://doi.org/10.1073/pnas.1717342115 513 Yang, L. H., & Rudolf, V. H. W. (2010). Phenology, ontogeny and the effects of climate change on the 514 timing of species interactions. Ecology Letters, 13(1), 1-10. https://doi.org/10.1111/j.1461- 515 0248.2009.01402.x 516 Zohner, C. M., Benito, B. M., Fridley, J. D., Svenning, J.‐C. & Renner, S. S. (2017) Spring predictability 517 explains different leaf-out strategies in the woody floras of North America, Europe and East Asia. 518 Ecology Letters, 20(4): 452-460. https://doi.org/0.1111/ele.12746 519 Zohner, C. M., Benito, B. M., Svenning, J.‐ C., & Renner, S. S. (2016). Day length unlikely to constrain 520 climate-driven shifts in leaf-out times of northern woody plants. Nature Climate Change, 6, 1120– 521 1123. https://doi.org/10.1038/nclimate3138 522 Zohner, C. M., Mo, L., Pugh, T. A. M., Bastin, J.-F., & Crowther, T. W. (2020). Interactive climate 523 factors restrict future increases in spring productivity of temperate and boreal trees. Global Change 524 Biology, 00, 1-14. https://doi.org/10.1111/gcb.15098 525 Zohner, C. M., Mo, L., & Renner, S. S. (2018). Global warming reduces leaf-out and flowering 526 synchrony among individuals. eLife, 7, e40214. https://doi.org/10.7554/eLife.40214 22 527 Figure legends 528 Fig. 1 Distribution of mean leaf flushing dates in early and late flushing species 529 (EFS and LFS, respectively) and changes in leaf flushing orders. (a) Distribution 530 of leaf flushing for each species during the period 1951−2016. (b) Direction and 531 magnitude of phenological shift (indicated by arrow direction and length) in leaf 532 flushing dates of EFS and LFS. (c) Direction and proportion of changes in leaf flushing 533 orders across 305 sites within EFS and LFS. The arrow indicates the direction of 534 changes in leaf flushing orders among species within group. Each value in the tables 535 indicate the percentage of sites where the leaf flushing order of the two species in the 536 corresponding row and column is interchanged. BP: Betula pendula; AH: Aesculus 537 hippocastanum; AG: Alnus glutinosa; FS: Fagus sylvatica; QR: Quercus robur; and, 538 FE: Fraxinus excelsior. 539 Fig. 2 Temporal changes of temperature and leaf flushing sensitivity to 540 temperature (ST). (a) Mean monthly warming rates across all sites during the period 541 1951−2016. Mean monthly warming rate was calculated using linear regression 542 between mean monthly temperature and year. (b) Mean spring temperature during the 543 TRP (Spring Tmean) over the periods 1951−1983 and 1984−2016 for early and late 544 flushing species (EFS and LFS, respectively). (c) Changes of spring temperature 545 variance (TSD) during the TRP over the periods 1951-1983 and 1984-2016 for EFS and 546 LFS. (d) Temperature sensitivity of leaf flushing (ST) for EFS and LFS between 547 1951−1983 and 1984−2016, determined by reduced major-axis regression. Asterisks 548 indicate differences at P < 0.05. 23 549 Fig. 3 Changes in chilling and heat accumulation, as well as standard deviation of 550 leaf flushing dates (LOSD) between 1951−1983 and 1984−2016 in early and late 551 flushing species (EFS and LFS). (a) Difference in chilling accumulation between EFS 552 and LFS, which was calculated as the number of days when mean air temperature was 553 <5 °C from 1st November until leaf flushing date. (b) Difference in growing degree 554 days (GDD) between EFS and LFS, which was calculated as the sum of mean air 555 temperatures >5 °C from 1st January until leaf flushing date. Numbers in blue and red 556 font are means ± SE for 1951−1983 and 1984−2016, respectively. (c) Difference in 557 LOSD between 1951-1983 and 1984-2016 for EFS and LFS. Asterisks indicate 558 differences at P < 0.05. 559 Fig. 4 The interspecific differences in leaf flushing date (IDSP) between early and 560 late flushing species (EFS and LFS) varied along latitudes. All the sites were 561 averaged every 0.5 degree along the latitude gradient. 24 562 Fig. 1 25 563 Fig. 2 26 564 Fig. 3 27 565 Fig. 4 28