Response patterns of xylem and leaf phenology to temperature at the southwestern distribution boundary of Quercus robur: A multi-spatial study

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Response patterns of xylem and leaf phenology to temperature at the southwestern distribution boundary of Quercus robur: A multi-spatial study Authors: Guillermo Guada1, Rosa Ana Vázquez-Ruiz1, Ignacio García-González1 1Departamento de Botánica, Universidade de Santiago de Compostela, Campus Terra, Escola Politécnica Superior de Enxeñaría, Lugo 27002, Spain. Author for correspondence: Guillermo Guada Prada ORCID iD: 0000-0001-9579-543X Escola Politécnica Superior de Enxeñaría Universidade de Santiago de Compostela Campus Terra, 27002 Lugo (Spain) Tlf:+34 982822491 E-mail: [email protected] Title page 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Abstract We investigated how temperature patterns affect cambial activity and leaf phenology of oak across a wide range of natural woodlands at its southwestern distribution boundary. Understanding the climatic control of wood formation in dominant species is very relevant to infer tree responses to ongoing environmental changes and their impact on the carbon cycle. We selected nine sites along two elevation gradients from the coastline in northwestern Iberia, and sampled ten trees per site biweekly during 2012 and 2013. Leaf and cambial phenological phases were related to mean air temperature for 10-60 day running periods along the year to identify the most relevant time windows for cambium and leaf phenophases, and the relationships among them. The first earlywood vessels expanded before the appearance of small leaves, and subsequently underwent maturation to meet water requirements for full leaf unfolding. The advance or delay of cambial reactivation and budburst varied among sites and years modulated by spring temperature, and were respectively maximized by maximum and minimum values. Temperature can modify the onset of early phenophases of primary and secondary growth differently, and also the synchronicity between them. However, the maturation of the first earlywood vessels is necessary to undergo full leaf extension. Keywords: budburst, cambium, earlywood, latewood, ring-porous wood, xylogenesis. 1 1.Introduction 1 Quercus robur L. is a nemoral tree species widely distributed under different climatic 2 conditions throughout Europe (Gilliam, 2016), from the Baltic Sea to its southern 3 distribution boundary in the Iberian Peninsula. Along this boundary, the transition to the 4 Mediterranean region involves its progressive replacement by more drought-tolerant 5 species (Sánchez de Dios et al., 2009), first nemoral oaks, and evergreen oaks under 6 more limiting conditions. 7 One of these transitional areas is located in northwestern Iberian, where prevailing 8 Atlantic conditions involve high precipitation records mainly during autumn and winter, 9 but summer drought can often occur as a consequence of Mediterranean influence. In 10 coastal areas, winter temperature is not limiting, but warm summer conditions facilitate 11 the occurrence of thermophile species, or even subtropical elements where summer 12 drought is not limiting (Izco et al., 1990). As a result, trees need to couple their patterns 13 of phenology and wood formation to these meteorological conditions. 14 Activity of primary and secondary meristems follows a distinct annual pattern in crown 15 and cambium, respectively. These processes are constraint to a well-defined time period 16 from spring to autumn, when conditions are favorable for growth and reproduction, 17 whereas trees enter dormancy during winter. Spatial and temporal changes in 18 environmental conditions can modify tree phenology (Menzel et al., 2006; Vitasse et al., 19 2017), as species adjust the timing and length of their growth and reproduction phases 20 to climate, with variations that depend on specific regional drivers, local adaptations, or 21 individual plasticity (Rossi et al., 2013). Though also interacting with photoperiod 22 (Basler and Korner, 2014), phenological events related to growth resumption are mainly 23 driven by temperature (Begum et al., 2013; Rossi et al., 2016), and consequently 24 changes related to the ongoing global warming have been observed (Menzel et al., 25 2006). 26 Q. robur, as a deciduous ring-porous hardwood species, needs to yearly renew its xylem 27 vessel network, so that the cambium must build the mechanical and conductive support 28 to fulfill water requirements from roots to leaves. The main pathways of sap flow are 29 the networks of earlywood vessels of the current year, because those from the previous 30 season are no more functional after a year (Chaney and Kozlowski, 1977; Ellmore and 31 2 Ewers, 1986; Umebayashi et al., 2008). The differentiation of current year’s earlywood 32 vessels has been detected to take place before the onset of bud growth (Lavrič et al., 33 2017; Pérez-de-Lis et al., 2016; Puchałka et al., 2017; Sass-Klaassen et al., 2011; 34 Takahashi et al., 2015). However, the importance of their contribution for crown 35 development is still under discussion (Kudo et al., 2018). Recent studies on the specific 36 moment when the networks of vessel elements become functional as conduits for water 37 movement suggest that the requirements of the early stages of leaf expansion can still be 38 fulfilled by latewood vessels formed in previous years (Kitin and Funada, 2016; Kudo 39 et al., 2015). Therefore, it is necessary to understand not only the mechanisms of wood 40 formation, but also physiological aspects of the tree growth. 41 Variations in the meteorological conditions prior or/and at the moment of specific 42 phenological events can affect the synchronization between primary and secondary 43 growth in a different way. Buds, leaves, and flowers that grow from primary meristems 44 are easily visible to the naked eye, and numerous records allow assessing variations in 45 the timing of leaf phenology. For example, Rossi (2015) found that temperature is a 46 predominant factor driving the ecotypic differentiation of budburst in black spruce. 47 Recent studies on different oak provenances in England (Wilkinson et al., 2017) 48 reported an advance in spring budburst across a range of temperatures, and Čufar et al. 49 (2012) found that March and April temperatures were related leaf unfolding in beech 50 (Fagus sylvatica) in Slovenia. During the growing season, vascular cambium cannot be 51 directly observed to record phenological events, but in the last decades, a great effort 52 has been made to identify stages of wood formation at a cellular level, although this 53 monitoring of xylogenesis is a very time-consuming method. 54 In conifers, linear and nonlinear patterns of timings and duration of wood phenology 55 showed local annual temperature as the main driver of cambial activity in the northern 56 hemisphere (Rossi et al., 2016; Rossi et al., 2013). Similarly, a phenological advance in 57 xylem phenology linked to temperature was found to modify the production of wood 58 biomass in northeastern France (Cuny et al., 2015). However, the study of xylogenesis 59 in hardwoods has deserved less attention. In the case of ring-porous oaks (Q. robur and 60 Q. pyrenaica), Pérez-de-Lis et al. (2017) demonstrated that the length of the growing 61 season was modulated by predisposing the number of dormant cambium cells, whereas 62 the size of the first earlywood vessels was affected by the timing of earlywood 63 enlargement (Pérez-de-Lis et al., 2016). The reactivation of cambial cells by heat 64 3 treatment was also proven for deciduous species (Begum et al., 2013; Kudo et al., 65 2014); similarly, differences in the yearly onset can be explained by temperature 66 variations (Prislan et al., 2013). 67 Although there is much evidence that temperature is one of the main driving forces for 68 plant growth in terrestrial ecosystems, its influence on wood formation has been mainly 69 analyzed on conifers, often at high latitudes or elevations (Lorena et al., 2016; Rossi et 70 al., 2016), whereas studies on cambial dynamics of temperate hardwood species such as 71 oaks are limited to very few sites (Lavrič et al., 2017; Pérez-de-Lis et al., 2017; 72 Puchałka et al., 2017). On the other hand, several studies dealt with the variation of 73 phenology across gradients, but were not linked to secondary growth (Vitasse et al., 74 2017). In order to fill this gap, we intend to evaluate the influence of the temperature 75 regime on the primary and secondary growth of native Q. robur forests within a region 76 that covers broad environmental gradients. For this, we compiled a data set of cambial 77 dynamics and leaf phenological phases at nine sites during 2012 and 2013, and related 78 their different phases to temperature. These sites were located towards the most 79 southwestern distribution boundary of oak, and represent Atlantic and Cantabrian 80 influences, from mild coastal areas to high continental inland, which to our knowledge 81 provides the widest range of these data currently available for oak. The aim of our study 82 is to establish the role of temperature on wood formation and crown development by i) 83 comparing the timings of cambial activity and leaf phenological phases, and ii) 84 evaluating how these relationships vary along different microclimatic gradients. 85 2. Material and Methods 86 2.1 Study site and tree selection 87 The study was conducted at nine Q. robur stands in northwestern Iberia (Table 1). The 88 selected sites covered the course of two rivers, Eume (E) and Sor (S), from their upper 89 watershed to their mouths into the Atlantic Ocean or the Cantabrian Sea respectively 90 (Fig. 1a). The area presents a mild Atlantic climate, with a mean annual temperature 91 ranging 9-13 ºC among sites, and a total precipitation of 900-1,500 mm; maximum 92 rainfall occurs during autumn-winter, and there is a varying degree of summer drought 93 depending on elevation or position towards the coast. Oak forests in this area are 94 4 characterized by moist and warm conditions, which even result in the occurrence of 95 some subtropical plants (Izco et al., 1990), especially at low altitude. 96 Within each watershed, we followed an altitudinal gradient that covered the whole 97 species distribution, with increasing degrees of precipitation with elevation, and 98 continentality from the coast to the inland mountains. Nevertheless, these two rivers 99 represent two dominant climatic influences on oak formations in northwestern Iberia. 100 Whereas the Atlantic coast is dominated by a certain Mediterranean trend during 101 summer (higher temperature, more intense drought), the Cantabrian coast is exposed to 102 mild northern winds that increase humidity and attemperate maximum temperature 103 (Martínez Cortizas and Pérez Alberti, 1999). 104 We monitored these forests for two consecutive years, from March to November in 105 2012 (1,175 trees sampled), and from February to September in 2013 (1,000 trees 106 sampled). For this, 90 trees (9 locations × 10 trees) with a diameter of 20-40 cm were 107 randomly selected at each sampling date within an area of ca. 2 ha per forest. The 108 sampling interval was 12-15 days in spring, and 20 days in summer, because changes 109 during earlywood formation or transition to latewood take place much faster than later 110 in the season. Trees selected were dominant or codominant in the canopy, and 111 individuals with polycormic stems, partially dead crowns, or evident damage, were 112 avoided. The random selection of trees at each date prevented the study at the individual 113 tree level, but provided a representative sample of the whole forest as a single 114 population, avoiding the bias of a specific tree. In addition, this strategy increased the 115 number of individuals that could be analyzed to relate primary and secondary growth, 116 i.e., to associate leaf and cambium phenophases. 117 2.2 Site meteorological data 118 Air temperature was hourly monitored at each site during the sampling period using 119 sensors (iButton DS1922L, San Jose, CA, USA) set at 2 m above the ground surface. 120 We calculated daily mean, maximum, and minimum values, as well as the 121 corresponding thermal amplitudes. Temperature time series were corrected to complete 122 missing data by simple linear regression with daily records from nearby meteorological 123 stations belonging to the weather service Meteogalicia (http://www.meteogalicia.gal/). 124 We compared all study sites according to their monthly mean temperatures for both 125 5 study years using hierarchical cluster analysis. The Euclidean distance was selected to 126 calculate dissimilarities among sites, whereas the average linkage was used as grouping 127 method (Fig. 1b). 128 2.3 Xylem sampling and wood phenophases 129 At each sampling date, a minimum of two microcores (2 mm in diameter, 15 mm in 130 length) was extracted at breast height (1.3 m) out of each selected tree by means of a 131 Trephor tool (Rossi et al., 2006), in perpendicular direction to the slope. Samples 132 contained mature and developing xylem of the current year, the cambial zone and 133 adjacent phloem, and at least one previous complete tree ring. After extraction, one 134 microcore per tree was processed by embedding in paraffin, cutting into thin sections 135 with a rotary microtome, and staining following the same protocol as in previous works 136 (Guada et al., 2018; Pérez-de-Lis et al., 2016). 137 Observation and width measurements of the wood phenophases of the 2,173 microcore 138 cross-sections were performed on images taken with a digital camera (Canon EOS 139 600D, Tokyo, Japan), coupled to a transmitted light microscope (Olympus BX40, 140 Tokyo, Japan); a white light polarizing filter allowed the detection of secondary cell 141 wall deposition (40 × magnification). We measured the width of cell expansion and 142 maturation zones along three radial lines per image (Fig. S1); and each phase of cambial 143 activity was expressed as particular day of year (DOY), including: beginning of cell 144 enlargement in the earlywood (bE), beginning of earlywood maturation (bM), beginning 145 of latewood maturation (bLW), and cessation of cell expansion (cE). The beginning of 146 cell enlargement was considered when the most recent ring contained at least one 147 enlarging cell; whereas the beginning of earlywood vessel maturation was defined by 148 the deposition of the secondary cell wall in the vessels, detected by birefringence under 149 polarizing light; the cessation of cell expansion corresponded to the moment when no 150 expanding cells were further detected, and the beginning of latewood maturation was 151 defined as the moment when earlywood growth had already ceased, and maturation was 152 detected in cells formed beyond the earlywood. 153 The durations between the onset and cessation of these phases were also calculated for 154 cell expansion (dE = cE − bE), earlywood maturation (dEW = bLW − bM), latewood 155 6 maturation (dLW = cE − bLW), and cell expansion of first row of earlywood vessels 156 (d1r = bM − bE). 157 2.4 Leaf phenological observations 158 Leaf phenology of each sampled tree was recorded in the field. Observations were 159 performed at the upper main branches using binoculars (10 × magnification), and 160 expressed as DOY. 161 We considered four different phenophases, namely budburst (BB), leaf unfolding (LU), 162 appearance of small leaves (SL; leaves < 50% of their apparent final size), and full 163 extension (FL; leaves > 50% of their apparent final size). Bud dormancy corresponded 164 to the overwintering stage, while bud swelling was identified by the apparent separation 165 of buds from the stem, together with the exposition of areas of lighter colored tissue as a 166 result of the initial extension of cataphylls with the separation of scales. Budburst was 167 characterized by green-colored expanded buds with no unfolded leaves, and leaf 168 unfolding ended as soon as the leaf blade was clearly visible, but not the petiole. 169 Appearance of small leaves was defined as the moment when at least one leaf was 170 completely out of the bud, and current year twigs and petiole could be visually 171 appreciated; and full extension was recorded when leaves attained at least the 50% of 172 their apparent final size. 173 2.5 Statistical analyses 174 In order to assess the critical dates of xylem and leaf phenology, the information on the 175 absence or presence of each phenological event per site and year was expressed as 176 binary data (0 no active phenophase; 1 active phenophase). Afterwards, we performed 177 logistic regressions by GLM for each phenophase using the DOY as independent 178 variable, and considering a probability of 0.5 as the most likely DOY for the activity of 179 a given event (Rathgeber et al., 2011). The effects of year, site, and their interaction 180 were studied by comparing the best logistic regression fit when these factors were 181 included isolated or in combination as classification variables. Independent variables 182 considered in the models tested were as follows: DOY + year; DOY+ site; DOY + year 183 + site; DOY + year × site. The model providing the lowest corrected Akaike's 184 Information Criterion (AICc) for each event was selected as the best fit, and a multiple 185 comparison between years and sites was performed using the Tukey test. 186 7 Relationships among the dates of wood phenological events estimated from the logistic 187 regression and their durations were related using Pearson’s correlations. The DOYs of 188 cessation of cell expansion among sites diverged significantly from normality, and were 189 transformed using the Y2 formula. 190 We used two approaches to establish relationships between primary and secondary 191 growth at tree scale, and specifically to look for a possible link between vessel 192 formation and budbreak. First, we selected the trees achieving the expansion of first row 193 of earlywood vessels, i.e., those showing enlarging vessels adjacent to the previous 194 annual ring, but still no maturation, and classified them according to their leaf 195 phenology in order to obtain the proportion of trees that started crown development 196 before earlywood maturation. The second approach consisted of selecting the trees with 197 already mature earlywood vessels for each phenological crown phase, and consequently 198 we established the proportion of trees that had started earlywood maturation for each 199 stage of leaf development. 200 In order to detect the time window that influenced primary and secondary growth, we 201 obtained moving averages of climatic data, and compared them to the DOYs of each 202 phenological event along the growing season. Thus, we preliminary used 30-day 203 averages of maximum and minimum temperatures, as well as thermal amplitude, and 204 correlated them to each phenological event every 15 days. A more detailed analysis of 205 the most relevant variables involved running temperature means of 10 to 60 days, 206 shifted in only one day. For these analyses, we assumed that the increase of temperature 207 advanced along with tree growth, and performed a linear model between the 208 temperature before the event and the timing of each event per site, including the year as 209 class variable. For each linear model, the p-value of a Type I test, and the corresponding 210 percentage of variation explained, were provided for the temperature effect and for the 211 difference between years. We also estimated temperature thresholds from those in the 212 DOY predicted by the GLM logistic curve per site and year. All statistical analyses 213 were performed using R statistical software (R Core Team, 2017). 214 3. Results 215 3.1 Temperature across sites 216 14 4.2 Primary and secondary growth relationships 401 There is an increasing interest in the relationships between primary and secondary 402 growth in ring-porous trees (e.g., Sass-Klaassen et al. (2011)), in order to understand the 403 influence of climate on the development of earlywood vessels, because it has been 404 shown to be a powerful dendroclimatic proxy (Souto-Herrero et al., 2017). This linkage 405 between crown phenology and intra-annual dynamics of xylem formation has been 406 investigated for different ring-porous hardwood species, as summarized by Kitin and 407 Funada (2016). According to the literature, the enlargement of the first vessel elements 408 starts before budburst, but their maturation onset differs among studies (Kudo et al., 409 2015; Pérez-de-Lis et al., 2016; Puchałka et al., 2017); in fact, Puchałka et al. (2017) 410 attributed such differences to genetic variability. But the number of individuals was 411 very limited in previous studies; in contrast, our study based on a sample size of 750 412 trees randomly selected during spring in two years. We observed that only 38-46% of 413 the individuals had started earlywood vessel maturation at the moment of leaf 414 unfolding, and even most of the trees with small leaves (76-85%) had already 415 undergone the maturation of the first vessel row. In addition, out of a sample of 314 416 trees with earlywood cell enlargement and still no secondary wall, more than the half 417 had not achieved budburst yet (75% in 2012, 56% in 2013), and a small proportion 418 (14% in 2012, 22% in 2013) was at the stage of budburst; later leaf phenophases were 419 hardly detected at this stage of vessel enlargement. 420 In view of these findings, we hypothesize that the newly-formed vessels are not capable 421 of supplying enough water for bud swelling and leaf unfolding in most trees yet. In this 422 sense, Pérez-de-Lis et al. (2016) also reported that secondary wall deposition was not 423 initiated at the moment of budburst in some Q. robur trees, Kudo et al. (2015) found 424 that the first earlywood vessels were completed along the entire stem only when small 425 leaves were visible to the naked eye in Q. serrata; and Guada et al. (2018) described the 426 maturation of the first vessel row to be concomitant with the leaf phenophases having 427 the greatest demand for water, such as leaf unfolding and extension. Kitin and Funada 428 (2016) concluded that water requirements for the initial leaf development needed to be 429 fulfilled by latewood vessels formed in previous years. Therefore, these latewood 430 vessels are probably enough to provide water before the appearance of small leaves, as 431 requirements at this stage are minimal in contrast to the amount of water needed for full 432 leaf expansion, which should require new functional earlywood vessels for more water 433 15 conduction (Lavrič et al., 2017). Thus, Guada et al. (2018) quantified the water content 434 of different leaf phenophases in Q. pyrenaica, and found that water transport of newly435 formed earlywood vessels was required to maintain the turgidity of extending leaves 436 and stems, but was not fundamental for the earliest phenological stages (bud swelling 437 and leaf unfolding). 438 4.3 Phenology and temperature 439 The relationship between cambial activity and leaf development previously described 440 help us understand the role of new earlywood vessels on the successful completion of 441 crown development in ring-porous hardwoods. However, this internal relationship can 442 be modified by external factors such as climatic conditions, which can somehow lead to 443 advances or delays in the different phenological events. 444 Spring temperature is the major driving force for cambium and crown reactivation, but 445 most investigations on its relation to xylogenesis were based on conifers at high 446 latitudes or elevations (Rossi et al., 2016; Rossi et al., 2008). Our results based on a 447 temperate ring-porous hardwood suggest that resumption of primary growth requires a 448 minimum threshold temperature, which is probably related to the need of a certain 449 thermal accumulation (Wilkinson et al., 2017). On the other hand, expansion of cambial 450 derivatives is more dependent on concomitant warm temperatures. The fact that 451 budburst is more linked to minimum temperatures while cell enlargement relies on 452 maximum values causes that the process of vessel maturation can take place slightly 453 before or after leaf unfolding depending on the conditions of each specific year. In fact, 454 we estimate that minimal temperature from mid-March to mid-April increases budburst 455 by 7.2 DOYs ºC-1, while March maximum temperature anticipates cell enlargement by 456 2.8 DOYs ºC-1, so that the variations between both processes are not coupled. However, 457 the new functional earlywood vessels are required for full shoot and leaf expansion 458 (Guada et al., 2018). 459 The influence of temperature on cambium reactivation has also been confirmed for 460 other deciduous hardwoods. Thus, Kudo et al. (2014) showed that localized heating for 461 six weeks induced earlier cambial reactivation in seedlings of Q. serrata, while a similar 462 result was found for a 4-week heating in poplar (P. sieboldii x P. grandidentata) 463 (Begum et al., 2007). In both studies, elevated temperature (20 ± 5 ºC) caused earlier 464 16 xylem differentiation, suggesting that the variation of temperature modifies the onset of 465 cell enlargement. Our study evidenced that the maximum temperature and thermal 466 amplitude (i.e., the difference between maximum and minimum daily temperatures) 467 anticipated cell enlargement in 2012 compared to 2013, as shown by the average of 90 468 trees. In contrast, the minimum temperature was not able to explain the differences in 469 cell enlargement between years. According to Begum et al. (2010), cambial reactivation 470 can be predicted considering the daily maximum temperature; likewise, we are the 471 opinion that an increasing temperature (and thermal amplitude) is an important factor 472 for cambial differentiation of the first vessels elements in the ring-porous hardwood Q. 473 robur. 474 Although crown development has been reported to require a minimum threshold to 475 break dormancy (Caffarra and Donnelly, 2011; Prislan et al., 2013; Wilkinson et al., 476 2017), budburst is considered to be also induced by photoperiod (Basler and Korner, 477 2014). Yet, photoperiod differences are small in our study due to the latitudinal 478 proximity among all sites; as a result, minimum spring temperature appears to be the 479 main factor affecting budburst of Q. robur in this area. This is consistent with the results 480 reported by Wilkinson et al. (2017) for the south of England, who found a negative 481 correlation between the date of budburst and the mean daily air temperature for Q. 482 robur and Q. petraea, whereby spring warming had a considerably larger effect on 483 budburst than winter chilling. 484 While wood formation responds to temperature increments probably by increasing its 485 growth rate, it is the variation in the minimum and maximum temperatures that 486 advances or delays the onset of primary and secondary growth differently across years 487 and sites (Fig. 6a). The broad sampling strategy involving nine sites and two study 488 years allowed us to estimate the average threshold temperature for each event (Fig. 6b). 489 The different timing for the beginning of cell expansion can be explained by the 490 considerable variation in maximum temperature between years; but this is not the case 491 for leaf unfolding, which hardly varied between 2012 and 2013, probably due to the 492 similar values of minimum temperature. 493 4.4 Conclusions 494 17 Primary and secondary growth of Q. robur at its southwestern distribution boundary 495 was closely linked to the moment of budburst and vessel enlargement. Practically all 496 trees began earlywood vessel enlargement before budburst, and only when at least the 497 first new vessels had completed maturation, did current leaves undergo full expansion. 498 Therefore, previous year’s vessels are enough to fulfill water requirements for budburst 499 and initial leaf growth, but current year’s vessels are needed to provide enough water 500 for further leaf and shoot development. 501 Differences in spring (or late winter) temperature regime between years and among sites 502 were able to explain the advancement and delay of leaf and earlywood development. 503 However, shoot and cambium activity were not controlled by the same environmental 504 conditions, because minimum temperature determined budburst, whereas maximum 505 temperature was responsible for the onset of vessel formation. Consequently, 506 understanding the environmental control of growth resumption in Q. robur is not 507 straightforward, because not only the timings, but also the relationship between primary 508 and secondary growth, can be modified by external factors. 509 5. Acknowledgments 510 This study was supported by the Spanish Ministry of Science and Innovation (Research 511 Project BFU-2010-21451) and Xunta de Galicia (Research Project ROCLIGAL, 512 10MDS291009PR). G. Guada acknowledges a predoctoral fellowship from FPI 513 program (BES-2011-050172) by the Spanish Ministry of Economy and Competitivity. I. 514 García-González and Rosa Ana Vázquez-Ruiz are grateful to Xunta de Galicia for the 515 recognition of Competitive Reference Group (2015/008, GI-1809-BIOAPLIC). The 516 authors are also grateful to L. Costa and C. Franco from ‘Fragas do Eume’ Natural Park 517 (Xunta de Galicia), M. Souto-Herrero for mapping, and G. Pérez-de-Lis for field 518 assistance. The frame of the COST Action FP1106 ‘STReESS’ inspired this research. 519 6. Author Contribution 520 I.G-G and G.G. planned and designed the research. G.G. performed experiments, 521 conducted fieldwork, and analysed data. G.G, I.G-G., and R. V-R interpreted the results. 522 G.G. and I.G-G. wrote the manuscript, with inputs from R.V-R. 523 524 18 Tables 525 Table 1 Description of the sites included in the analysis, with their identification codes 526 (ID), the corresponding river catchment (Eume, Atlantic; and Sor, Cantabrian), site 527 name, geographical location, elevation, and diameter at breast height (DBH) of the 528 sampled trees. Annual temperature (Temp) and precipitation (Prec) values obtained 529 from Rodríguez-Lado et al. (2016) (source: digital database at www.rgis.cesga.es). 530 ID River Latitude (N) Longitude (E) Elevation (m a s l) DBH (cm) Temp (ºC) Prec (mm) E1 Eume 43.416991 -8.064395 125 29.1 ± 4.5 13.2 1158 E2 Eume 43.372757 -7.991181 350 29.0 ± 5.4 12.2 1268 E3 Eume 43.469576 -7.788688 450 29.3 ± 4.4 12.0 1427 E4 Eume 43.448001 -7.634295 600 30.0 ± 4.8 9.5 1526 S1 Sor 43.676000 -7.708347 125 29.5 ± 5.2 13.0 1194 S2 Sor 43.594782 -7.721225 275 28.0 ± 5.4 12.9 1232 S3 Sor 43.556076 -7.740287 350 30.0 ± 5.8 12.4 1346 S4 Sor 43.529401 -7.736388 500 24.8 ± 3.9 12.4 1361 S5 Sor 43.495749 -7.739470 625 25.7 ± 3.6 11.0 1442 531 532 19 Table 2 Pearson’s correlation matrix between the date of cambium phenology events 533 (DOY) and the duration of wood maturation phases (days). The events are: beginning of 534 cell enlargement (bE); earlywood maturation (bM); latewood maturation (bLW); and 535 cessation of cell expansion (cE with a normal-transformation Y2). The durations are 536 referred to cell expansion (dE), earlywood maturation (dEW), latewood maturation 537 (dLW), and cell expansion of first row of earlywood vessels (d1r). 538 bE cE bM bLW dE dEW dLW d1r bE < 0.001 < 0.05 < 0.05 cE 0.170 < 0.001 < 0.001 bM 0.821 0.358 bLW 0.560 -0.022 0.562 < 0.01 < 0.01 dE -0.478 0.791 -0.197 -0.357 < 0.001 dEW -0.086 -0.349 -0.244 0.665 -0.240 < 0.01 dLW -0.270 0.726 -0.138 -0.703 0.808 -0.700 d1r -0.192 0.361 0.402 0.069 0.428 -0.282 0.197 539 540 20 Table 3 Generalized linear models to evaluate the effect of site and year on the 541 occurrence of the main phenological events. AIC increments (DAICc) for each model 542 are shown as compared to those of the model with the lowest score (the best-fitted 543 model). Selected models are highlighted in bold. Beginning of cell enlargement (bE); 544 beginning of maturation (bM); beginning of latewood maturation (bLW); cessation of 545 cell expansion (cE); budburst (BB); leaf unfolding (LU); <50% of final leaf size (SL); 546 >50% of final leaf size (FL); d.f. refers to degrees of freedom. 547 Xylogenesis Leaf phenology Fixed effects d.f. bE bM bLW cE BB LU SL FL DOY + Site x Year 19 0 0 0 3.50 0 0 0 0 DOY + Site + Year 11 30.94 11.32 6.52 0 84.29 4.01 21.34 13.92 DOY + Site 10 66.06 19.69 52.80 5.98 106.54 5.05 19.45 23.63 DOY + Year 3 130.39 87.88 43.53 20.97 208.96 170.68 199.68 231.26 548 21 Table 4 Significant (P < 0.05) statistical values of the linear regression for the different 549 events, expressed as day of the year (DOY) or duration in days as dependent variable; 550 and the mean, amplitude (Ampl.) minimum (Min.), and maximum (Max.) site 551 temperature for the period of each event as independent variable; year is considered as 552 fixed factor. The values shown are: intercept (a); slope (b); residual standard error of the 553 regression (RSE); and adjusted R2 of the model, including temperature and year effects 554 (Adj. R2). The P-value of Type I test is provided for the temperature effect, as well as 555 the difference between years, expressed in days, whenever this effect was significant. 556 Variables with the highest R2 are shown in bold. 557 Events Period Temp. b a RSE Year effect Type I P Adj. R2 Beginning of cell enlargement (bE) March Mean 137.99 -5.09 6.68 ns < 0.001 0.5219 Max. 130.69 -2.82 5.99 ns < 0.001 0.6161 Ampl. 122.31 -3.90 6.24 ns < 0.001 0.5821 Budburst (BB) Mid-March to mid-April Mean 165.90 -6.40 8.94 -10 < 0.01 0.472 Min. 145.79 -7.22 8.06 ns < 0.001 0.567 Beginning of cell maturation (bM) March Mean 172.51 -5.54 6.90 ns < 0.001 0.555 Min. 147.78 -5.37 8.21 9 < 0.05 0.3719 Max. 190.16 -4.48 6.36 -12 < 0.001 0.6225 Duration of 1st vessels enlargement (d1r) April Mean 53.98 -2.72 4.79 ns < 0.01 0.367 Min. 46.62 -2.83 5.13 ns < 0.05 0.271 Max. 57.77 -2.20 4.52 ns < 0.01 0.434 Duration of latewood (dLW) October Max. -27.103 5.78 14.26 -23 0.05 0.45 558 22 Figures 559 560 Figure 1. Location of the study sites in the northwestern Iberian Peninsula, and 561 corresponding hierarchical classification following temperature regime. (a) Q. robur 562 distribution map (www.euforgen.org); and the study sites, Eume (E1-E4) and Sor (S1563 S5) river. (b) Dendrograms of monthly mean, maximum, and minimum temperatures 564 for the nine sites, clustered into three similar climatic conditions G1, G2, G3. 565 566 23 567 Figure 2 Relationship of primary and secondary growth at tree scale. (a) Percentage of 568 different crown phenophases at the moment of beginning of cell enlargement; number 569 of trees per each crown phenophase are indicated above the column. (b) Percentage of 570 trees with vessel maturation for each crown phenophase in spring; the number of trees 571 with maturation out of the total amount trees for each specific crown phenophase is 572 indicated above the column. No budburst (No BB), budburst (BB), leaf unfolding (LU), 573 small leaves (SL), full leaf expansion (FL). 574 575 30 Lorena, B. et al., 2016. Compensatory mechanisms mitigate the effect of warming and 676 drought on wood formation. Plant, Cell & Environment, 39(6): 1338-1352. 677 http://dx.doi.org/doi:10.1111/pce.12689. 678 Martínez Cortizas, A. and Pérez Alberti, A., 1999. Dominios ombrotérmicos. In: A. 679 Martínez Cortizas and A. Pérez Alberti (Eds.), Atlas climático de Galicia. Xunta 680 de Galicia, pp. 109-111. 681 Menzel, A. et al., 2006. European phenological response to climate change matches the 682 warming pattern. Global Change Biology, 12(10): 1969-1976. 683 http://dx.doi.org/10.1111/j.1365-2486.2006.01193.x. 684 Pérez-de-Lis, G. et al., 2017. Environmental conditions and vascular cambium regulate 685 carbon allocation to xylem growth in deciduous oaks. Functional Ecology, 686 31(3): 592-603. http://dx.doi.org/10.1111/1365-2435.12789. 687 Pérez-de-Lis, G., Rossi, S., Vázquez-Ruiz, R.A., Rozas, V. and García-González, I., 688 2016. Do changes in spring phenology affect earlywood vessels? Perspective 689 from the xylogenesis monitoring of two sympatric ring-porous oaks. New 690 Phytologist, 209(2): 521-530. http://dx.doi.org/10.1111/nph.13610. 691 Prislan, P., Gričar, J., de Luis, M., Smith, K.T. and Čufar, K., 2013. Phenological 692 variation in xylem and phloem formation in Fagus sylvatica from two 693 contrasting sites. Agricultural and Forest Meteorology, 180: 142-151. 694 http://dx.doi.org/10.1016/j.agrformet.2013.06.001. 695 Puchałka, R., Koprowski, M., Gričar, J. and Przybylak, R., 2017. Does tree-ring 696 formation follow leaf phenology in Pedunculate oak (Quercus robur L.)? 697 European Journal of Forest Research, 136(2): 259-268. 698 http://dx.doi.org/10.1007/s10342-017-1026-7. 699 Rathgeber, C.B.K., Longuetaud, F., Mothe, F., Cuny, H. and Le Moguedec, G., 2011. 700 Phenology of wood formation: Data processing, analysis and visualisation using 701 R (package CAVIAR). Dendrochronologia, 29(3): 139-149. 702 http://dx.doi.org/10.1016/j.dendro.2011.01.004. 703 Rodríguez-Lado, L., Tapia del Río, L., Rodríguez-Pérez, M., Martínez-Cortizas, A., 704 Macías, F., Abad, E., 2016. Atlas digital de propiedades de suelos de Galicia. 705 Universidade de Santiago de Compostela, Santiago de Compostela (Spain). 706 http://rgis.cesga.es. 707 31 Rossi, S., 2015. Local adaptations and climate change: converging sensitivity of bud 708 break in black spruce provenances. International Journal of Biometeorology, 709 59(7): 827-835. http://dx.doi.org/10.1007/s00484-014-0900-y. 710 Rossi, S. et al., 2016. Pattern of xylem phenology in conifers of cold ecosystems at the 711 northern hemisphere. Global Change Biology, 22: 3804–3813. 712 http://dx.doi.org/10.1111/gcb.13317. 713 Rossi, S. et al., 2013. A meta-analysis of cambium phenology and growth: linear and 714 non-linear patterns in conifers of the northern hemisphere. Annals of botany, 715 112(9): 1911-1920. http://dx.doi.org/10.1093/aob/mct243. 716 Rossi, S., Anfodillo, T. and Menardi, R., 2006. Trephor: A new tool for sampling 717 microcores from tree stems. Iawa Journal, 27(1): 89-97. http://dx.doi.org/ 718 10.1163/22941932-90000139 719 Rossi, S. et al., 2008. Critical temperatures for xylogenesis in conifers of cold climates. 720 Global Ecology and Biogeography, 17(6): 696-707. 721 http://dx.doi.org/10.1111/j.1466-8238.2008.00417.x. 722 Sánchez de Dios, R., Benito-Garzón, M. and Sainz-Ollero, H., 2009. Present and future 723 extension of the Iberian submediterranean territories as determined from the 724 distribution of marcescent oaks. Plant Ecology, 204(2): 189-205. 725 http://dx.doi.org/10.1007/s11258-009-9584-5. 726 Sass-Klaassen, U., Sabajo, C.R. and den Ouden, J., 2011. Vessel formation in relation to 727 leaf phenology in pedunculate oak and European ash. Dendrochronologia, 29(3): 728 171-175. http://dx.doi.org/10.1016/j.dendro.2011.01.002. 729 Souto-Herrero, M., Rozas, V. and García-González, I., 2017. A 481-year chronology of 730 oak earlywood vessels as an age-independent climatic proxy in NW Iberia. 731 Global and Planetary Change, 155: 20-28. 732 http://dx.doi.org/https://doi.org/10.1016/j.gloplacha.2017.06.003. 733 Takahashi, S., Okada, N. and Nobuchi, T., 2015. Relationship between vessel porosity 734 and leaf emergence pattern in ringand diffuse-porous deciduous trees in a 735 temperate hardwood forest. Botany, 93(1): 31-39. http://dx.doi.org/10.1139/cjb736 2014-0129. 737 Umebayashi, T. et al., 2008. Conducting Pathways in North Temperate Deciduous 738 Broadleaved Trees. Iawa Journal, 29(3): 247-263. 739 http://dx.doi.org/10.1163/22941932-90000184. 740 32 Vitasse, Y., Signarbieux, C. and Fu, Y.H., 2017. Global warming leads to more uniform 741 spring phenology across elevations. Proceedings of the National Academy of 742 Sciences, 115: 1004-1008. http://dx.doi.org/10.1073/pnas.1717342115. 743 Wilkinson, M., Eaton, E.L. and Morison, J.I.L., 2017. Variation in the date of budburst 744 in Quercus robur and Q-petraea across a range of provenances grown in 745 Southern England. European Journal of Forest Research, 136(1): 1-12. 746 http://dx.doi.org/10.1007/s10342-016-0998-z. 747 Supporting Information Click here to download Supplementary Interactive Plot Data (CSV): Supporting Information_Robur_AFM.docx