Tree vigour influences secondary growth but not responsiveness to climatic variability in Holm oak

1 Tree vigour influences secondary growth but not responsiveness to climatic 1 variability in Holm oak 2 3 Ana-Maria Hereşa*, Margot W. Kayebc, Elena Grandad, Raquel Benavidesa, Ana 4 Lázaro-Nogala, Alfredo Emilio Rubio-Casale, Fernando Valladaresaf, Jorge Curiel 5 Yustegh 6 a Departamento de Biogeografía y Cambio Global, Museo Nacional de Ciencias 7 Naturales (MNCN, CSIC), Serrano 115 dpdo, E-28006 Madrid, Spain 8 b Intercollege Graduate Degree Program in Ecology, Pennsylvania State University, 9 303 Forest Resources Building, University Park, PA 16802, USA 10 c Department of Ecosystem Science and Management, Pennsylvania State 11 University, University Park, PA, USA 12 d Instituto Pirenaico de Ecologia (IPE, CSIC), Avda. Montañana 1005, E-50059, 13 Zaragoza, Spain 14 e Departamento de Biología Vegetal y Ecología, Universidad de Sevilla, Ap. 1095, E-15 41080, Sevilla, Spain 16 f Área de Biodiversidad y Conservación, Universidad Rey Juan Carlos, Tulipán s/n, 17 E-28933 Móstoles, Spain18 g BC3 - Basque Centre for Climate Change, Scientific Campus of the University of 19 the Basque Country, 48940 Leioa, Spain 20 h IKERBASQUE, Basque Foundation for Science, María Díaz de Haro 3, 6 solairua 21 48013 Bilbao, Bizkaia, Spain 22 *Corresponding author: [email protected]; phone: +34 917452500 ext.23 980805 24 This document is the Accepted Manuscript version of a Published Work that appeared in final form in: Heres A.-M., Kaye M.W., Granda E., Benavides R., Lázaro-Nogal A., Rubio-Casal A.E., Valladares F., Curiel Yuste J. 2018. Tree vigour influences secondary growth but not responsiveness to climatic variability in Holm oak. DENDROCHRONOLOGIA. 49. 68-76. DOI (10.1016/j.dendro.2018.03.004). © 2018 Elsevier GmbH This manuscript version is made available under the CC-BY-NC-ND 3.0 license http://creativecommons.org/licenses/by-nc-nd/3.0/ 2 25 ABSTRACT 26 Many tree species from Mediterranean regions have started to show increased rates 27 of crown defoliation, reduced growth, and dieback associated with the increase in 28 temperatures and changes in the frequency and intensity of drought events 29 experienced during the last decades. In this regard, Quercus ilex L. subsp. ballota 30 [Desf.] (Holm oak), despite being a drought-tolerant species widely distributed in the 31 Mediterranean basin, it has recently started to show acute signs of decline, extended 32 areas from Spain being affected. However, few studies have assessed the role of 33 climatic variability (i.e., temperature, precipitation, and drought) on the decline and 34 resilience of Holm oak. Here, we measured secondary growth of seventy Holm oaks 35 from a coppice stand located in central Spain. Sampled trees had different stages of 36 decline, so they were classified into four vigour groups considering their crown foliar 37 lost: healthy (0%), low defoliated (< 25%), highly defoliated (25 to 70%), and dying 38 (70 to 100%). Our results showed that during the study period (1980 - 2009) the 39 highly defoliated and dying Holm oaks grew significantly less than their healthy and 40 low defoliated neighbours, suggesting permanent growth reduction in the less 41 vigorous individuals. Despite these differences, all four vigour groups showed similar 42 responses to climatic variations, especially during winter and late spring – early 43 summer seasons, and similar resilience after severe drought events, managing to 44 significantly recover to pre-drought growth rates after only two years. Our findings, 45 hence, illustrate that tree vigour influences secondary growth but not responsiveness 46 to climatic variability in Holm oak. Still, as reduced growth rates are frequently 47 associated with the process of tree mortality, we conclude that the less vigorous 48 3 Holm oaks might not be able to cope with future water stress conditions, leading to 49 increased mortality rates among this emblematic Mediterranean species. 50 51 KEYWORDS 52 Holm oak; drought; tree-rings; coppice; defoliation. 53 54 INTRODUCTION 55 Tree decline is currently one of the main worrisome and studied issues in forest 56 ecology worldwide (Allen et al., 2010; Carnicer et al., 2011), being mainly associated 57 with increasing temperatures and changes in precipitation intensity and frequency 58 (IPCC, 2014). In the last decades this phenomenon has affected all major forest 59 biome types, and thus many conifer and broadleaf species, both evergreen and 60 deciduous, independent of their sensitivity or tolerance to stressful climatic 61 conditions (e.g., droughts) (see Allen et al., 2010). Tree decline leads to low 62 ecosystem productivity, changes in species distribution, and altered forest 63 succession, finally affecting all forest ecosystem services (Anderegg et al., 2013a). 64 Thus, to minimize and prevent such changes, identifying the causes that underlie 65 forest vulnerability and decline (e.g., defoliation, growth reduction, dieback), and how 66 these ecosystems respond to changes in climatic conditions, have become of 67 upmost importance. 68 Tree rings have been recognized as good predictors of increased vulnerability, 69 decline, and stress-induced mortality events. For instance, studies like Hereş et al. 70 (2012) or Cailleret et al. (2017) have identified that long-term decreases in radial 71 growth rates precede tree mortality events. However, in Cailleret et al. (2017) the 72 authors also highlighted that mortality can be preceded by quick growth declines, or 73 4 even by increased growth rates, with different patterns varying among tree species 74 and drought tolerance-strategies. As a consequence, species-specific studies are 75 needed to better identify the causes that underlie forest vulnerability and decline. 76 Mediterranean regions, where water availability is the main limiting factor for plant 77 growth (Cherubini et al., 2003; Martínez-Vilalta et al., 2008), are considered to be 78 especially vulnerable to increased temperature and changes in the frequency and 79 intensity of drought events (Giorgi and Lionello, 2008; IPCC, 2014). In the Iberian 80 Peninsula, where predicted climatic conditions are already felt (Kovats et al., 2014), 81 many tree species have started to show alarming signs of decline and dieback in the 82 last decades (Peñuelas et al., 2000; Martínez-Vilalta and Piñol, 2002; Camarero et 83 al., 2015a; Hereş et al., 2012; Natalini et al., 2016). Affected tree species include 84 those that reach their southernmost distribution limits in the Iberian Peninsula (e.g., 85 Pinus sylvestris L.), and that are considered to be particularly vulnerable to 86 conditions induced by climate change (e.g., droughts), but also species well adapted 87 to dry conditions and widely distributed in the Iberian Peninsula such as the Holm 88 oak (Quercus ilex L. subsp. ballota [Desf.]). 89 Holm oak is an evergreen tree species, usually retaining its leaves up to three years 90 (Montserrat-Martí et al., 2009). Its distribution covers western Iberia and North Africa 91 (Rodá et al., 1999), dominating areas with continental Mediterranean conditions 92 (Blanco et al., 1997). Almost 60% of all Holm oak areas are located in Spain 93 (Corcuera et al., 2004), where this species forms natural forests or dehesa 94 ecosystems, playing thus an important role from an environmental and socio-95 economic point of view (Patón et al., 2009). From all areas covered by Holm oak in 96 Spain, it is estimated that almost 44% of them are coppice stands (see Corcuera et 97 al., 2004 and references therein). Holm oaks tolerate thermal stress and precipitation 98 5 variability (Gratani, 1996), conditions that are characteristic for the continental 99 Mediterranean climate, being considered a species that can endure both winter frost 100 and summer drought (Terradas and Savé, 1992; Martínez-Vilalta et al., 2002; 101 Montserrat-Martí et al., 2009). Despite the drought-tolerant strategy provided by 102 evergreenness and narrow vessels in the diffuse-to semi-ring-porous wood 103 (Corcuera et al., 2004), this species has recently started to show acute signs of 104 decline (i.e., wilting of leaves, defoliation, growth reduction, dieback, etc.), with 105 extensive affected areas (Brasier, 1996; Gea-Izquierdo et al., 2011; Navarro, 2011; 106 Natalini et al., 2016). Climatic stress (e.g., drought) plays an important role in the 107 decline of this tree species (Gea-Izquierdo et al., 2011; Granda et al., 2013; 108 Camarero et al., 2015b; Camarero et al., 2016; Natalini et al., 2016), as well as other 109 factors such as the pathogenic oomycete Phytophthora cinnamomi, and intense 110 management practices (see Natalini et al., 2016 and references therein). 111 The main objective of this study was to assess the role of climatic variability (i.e., 112 temperature, precipitation, and drought) on the decline and resilience of Holm oak 113 trees growing in a coppice stand located in central Spain. We used tree-ring 114 chronologies that provide retrospective precise temporal insights into the climate-115 growth relationships of woody plants (Fritts, 1976), and allow comparison on long-116 term growth rates between trees at different stages of decline. Specifically, we 117 studied Holm oak trees growing in a coppice stand that were classified into four 118 vigour groups depending on their level of defoliation (% of crown foliar lost; used 119 here as a proxy to define the level of decline of the studied trees), with the aim to: i) 120 study potential differences in historical growth rates between trees of contrasting 121 defoliation levels; ii) assess whether early signs of growth decline or abrupt growth 122 changes may be identified in less vigorous trees; iii) analyse the annual and 123 6 seasonal climatic influence on the growth of these Holm oaks growing in a coppice 124 stand; and iv) evaluate the response of these trees during and after severe drought 125 events. 126 127 MATERIALS and METHODS 128 Study area 129 The study area hosts Holm oaks at varying stages of defoliation (between 0 and 130 100% of crown foliar lost). Tree-ring samples and site data were collected from two 131 coppice stands separated by <5 km, situated between two localities from the centre 132 of the Iberian Peninsula (Spain, Community of Madrid): Chapinería (40°23’03.4”N 133 4°11’37.8”W, ≈ 650 m a.s.l.) and Navas del Rey (40°23'55.80"N 4°14'26.70"W, ≈ 675 134 m a.s.l.). Holm oaks from Chapinería and Navas del Rey were treated as a single 135 database as the two coppice stands are close, have similar abiotic and stand 136 conditions, and no significant growth differences were found between the trees 137 sampled in each of them (W = 1330, p = 0.433). The study area is characterized by a 138 low tree density (≈180 trees ha-1; Rodríguez et al., 2016). The overstory is dominated 139 by Holm oak trees, with scarce Juniperus oxycedrus (Sibth. and Sm) present, while 140 the understory is represented by shrubs (Retama sp., Lavandula sp., etc.) and 141 herbaceous species (Vulpia sp., Bromus sp., Xolantha sp., etc.) (Rodríguez et al., 142 2016). The study area has been traditionally used for pasturage, hunting and 143 logging, although in the last decades human use has been considerably reduced, 144 and no signs of recent logging are observed. The soils have a pH of 6, are sandy 145 upon fractured bedrock mainly formed by biotite granites (8% of estimated superficial 146 stoniness) (García-Angulo et al., unpublished data), and belong to the Cambisols 147 type group (Monturiol Rodríguez and Alcalá del Olmo Bobadilla, 1990). The 148 7 steepness of the terrain is low, with an estimated slope average of 8%. The climate 149 of the study area is continental Mediterranean, characterized by hot and dry 150 summers and cold winters (Cuatro Vientos meteorological station; Spanish State 151 Agency of Meteorology). 152 153 Field sampling and tree ring analysis 154 Tree-ring widths were used to reconstruct past growth rates (Fritts, 1976) of adult 155 Holm oak trees showing variable rates of crown defoliation. Seventy Holm oak trees 156 of similar age (years) (Table 1), with no signs of attack by biotic factors (e.g., insects, 157 fungi), were selected and classified into four vigour groups, according to their crown 158 foliar percentage lost: healthy (0% of defoliation), low defoliated (< 25% of 159 defoliation), highly defoliated (25 to 70% of defoliation), and dying (70 to 100% 160 defoliation). To do so, four 300 m long transects (two in Chapinería and two in Navas 161 del Rey), placed 50 m away from each other (i.e., within the same coppice stand), 162 were established. Holm oaks were sampled within 10 m of the central transect line, 163 trying to balance the number of trees for each vigour group, and their distribution 164 within the study area. To monitor the vigour condition of the seventy selected Holm 165 oaks, the visual estimation of their crown foliar percentage lost was recorded in 166 October 2010 (our reference date; see below), and again within the next two 167 following years (i.e., May 2011, April 2012, and October 2012). Crown foliar 168 percentage lost was evaluated at the whole tree level, and was always estimated by 169 the same observer for consistency. To estimate crown foliar percentage lost, every 170 sampled tree was compared to a reference tree with no defoliation. The four vigour 171 groups differed in their diameter at breast height (DBH) given that a uniform 172 selection during sampling was not possible (Table 1). Field sampling was conducted 173 8 in October 2010 and consisted of extracting at breast height (standard 1.3 m above 174 from soil) at least two radial wood cores from the largest stem of each of the seventy 175 Holm oaks, using increment borers (5 mm diameter; Haglöf, Sweden). As the 176 outermost tree ring (i.e., 2010, the year of sampling) was not completely formed, 177 tree-ring chronologies ended in 2009. To avoid the juvenile effect that is 178 characterized by irregular tree rings in the first years of tree growth (Richter, 2015), 179 and to have sufficient sample depth per year for all Holm oak groups, we only 180 considered 1980-2009 for analyses (see below). 181 Following field sampling, cores were air-dried, glued and polished using a series of 182 sand-paper grits so tree-ring boundaries were clearly visible. Cores were first visually 183 cross-dated using wide and narrow pointer years (Stokes and Smiley, 1968) and 184 then measured to the nearest 0.01 mm using a LINTAB digital positioner and the 185 TSAP-WinTM software (Rinn, 2004). Cross-dating accuracy was repeatedly checked 186 using COFECHA (Holmes, 1983). In total 112 series from 70 trees were cross-dated, 187 with an inter-series correlation of 0.67 and an average mean sensitivity of 0.43. Our 188 intention was to analyse two radii per tree, but for 31 trees we could only cross-date 189 one radius: the second radius of 7 of them had branch scars or breaks that made 190 them not datable, while the second radius of the other 24 could not be dated due to 191 difficulties in distinguishing the ring boundaries, a common challenge for Holm oak 192 cross-dating (Gea-Izquierdo et al., 2009 and references therein) given its diffuse-to 193 semi-ring-porous wood with multiseriate rays (Schweingruber, 1990; Supplementary 194 Fig. 1). Of the 31 trees with one cross-dated radius, the cross-dated radius was 195 longer than the non-cross-dated one for 23 of the trees. We decided to use only one 196 cross-dated radius for all 70 Holm oaks, selecting for all of them the core with the 197 longer radius. To understand the potential bias introduced by analysing growth from 198 9 the longer radius only, we compared the average growth from the two radii to the 199 growth of the longer and found that the longer radii overestimated radial growth by 200 4%, and this pattern was consistent across tree diameters (linear regression 201 Radiuslonger (mm) = 1.04*Radiusaverage (mm) + 2.54; R2 = 0.95). The inter-series 202 correlation for the final selected 70 cores was 0.66 and the average mean sensitivity 203 had a value of 0.44. Our final chronology correlated well with a Holm oak chronology 204 obtained from individuals growing in a dehesa ecosystems situated in the Central 205 Mountain Range in Spain (R2 = 0.52, p < 0.001) (Gea-Izquierdo et al., 2009). 206 To remove the trend of decreasing ring width with increasing stem size and tree age 207 over time, and to have a better estimate of overall tree growth (Biondi and Qeadan, 208 2008), measured tree-ring widths were transformed into basal area increment (BAI) 209 using the dplR R package (Bunn, 2008; Bunn et al., 2016). In addition, tree-ring 210 widths were transformed into dimensionless ring width indices (RWI; needed for the 211 superposed epoch analyses (SEA), see below) with both age-related growth trends 212 and lower-frequency variation removed from the time series (Fritts, 1976; Cook and 213 Kairiukstis, 1990). For this, raw ring width data were double detrended: first, a 214 negative exponential curve or linear regression was fitted, and then a cubic 215 smoothing spline with a 50% frequency (Cook and Peters, 1981). These latter 216 analyses were done using the ARSTAN software (Cook and Holmes, 1986). 217 218 Climatic data 219 Mean monthly temperature (T, ºC) and accumulated precipitation (P, mm) data for 220 the 1973 - 2009 period were available from the Cuatro Vientos meteorological station 221 situated at ca. 36 km from the study area (data provided by the Spanish State 222 Agency of Meteorology). Average annual precipitation is 435 mm, with August being 223 16 spring - early summer (May [t] to June [t]) (p < 0.001) (Table 3). None of all these 373 relationships depended on the vigour group. For the seasonal P and SPEI models, 374 the least-squares means were always lower for the highly defoliated Holm oaks than 375 for the other three vigour groups, but significant differences were found only between 376 the highly and low defoliated ones (p < 0.05). 377 378 Growth response to severe drought events 379 SEA analyses revealed that all vigour groups responded similarly to severe drought 380 events (Fig. 3). Specifically, each of the four vigour groups of Holm oaks showed a 381 significant, sharp growth (RWI) reduction during severe drought events (year 0; p < 382 0.05): 58% for the healthy Holm oaks, and ca. 50% for the low defoliated, highly 383 defoliated, and dying ones. Two years after the severe drought events, all vigour 384 groups managed to significantly recover their growth rates (p < 0.05), registering 385 similar to pre-drought RWI values (Fig. 3). 386 387 DISCUSSION 388 Secondary growth of the drought-tolerant Holm oak varied between vigour groups 389 but not its responsiveness to climatic variability, nor its resilience following severe 390 drought events. Within the studied coppice stand, Holm oaks showing different 391 stages of decline cope with the increased aridity caused by warming coupled with no 392 parallel increases in rainfall (Supplementary Fig. 2b; Granda et al., 2013). As 393 expected, although none of the four vigour groups showed significant temporal 394 growth trends between 1980 and 2009, less vigorous Holm oaks (i.e., highly 395 defoliated and dying) grew significantly less than the most vigorous ones (i.e., 396 healthy and low defoliated). This pattern persisted during the whole study period. 397 17 Despite the lack of numeric records of Holm oak defoliation rates preceding the 398 sampling date (i.e., October 2010), the fact that the studied trees maintained the 399 same crown foliar percentage levels two years after, points out towards a synchrony 400 between growth rates (low or high) and crown condition (less or more vigorous). 401 Reduced growth rates systematically followed the severe drought events registered 402 during the study period as droughts reduce carbon allocation to stem growth through 403 stomatal closure, which eventually leads to crown defoliation, a common pattern 404 found for both conifer and broadleaf species (Dobbertin, 2005). According to 405 previous findings, crown defoliation following extreme droughts can prolong up to 406 seven years after these events (Galiano et al., 2012), which suggests that the 407 observed Holm oak defoliation could have occurred before 2010 given the repeated 408 severe drought events registered. Actually, the 2005 severe drought event resulted 409 in strong defoliation and mortality of Holm oaks in the studied coppice stand 410 (Valladares F., personal observation), which supports our hypothesis about the 411 presence of important defoliation rates before 2010. 412 Growth of Holm oak was similarly influenced by climatic variability regardless of the 413 crown condition. According to other results from Mediterranean forests (Natalini et 414 al., 2016) warmer winter temperatures enhanced growth of Holm oaks while high late 415 spring – early summer temperatures impaired it. Contrary to our results, Granda et 416 al. (2013) showed that Holm oak was favoured by high spring temperatures, 417 although our study site is more arid than the one considered in that study. 418 Interestingly, we did not find any significant climate-growth relationship for the early-419 spring (March-April, [t]), despite that Holm oak individuals usually register high 420 growth rates during this period (Corcuera et al., 2004; Gea-Izquierdo et al., 2009; 421 Granda et al., 2013; Natalini et al., 2016). This result could be explained by low early 422 18 spring precipitation, especially compared to those in May that almost doubled March 423 values (Supplementary Fig. 2a). On the other hand, higher precipitation and SPEI 424 favoured Holm oak growth, an effect that was more evident during the winter and 425 late spring – early summer periods. Spring and summer climatic conditions are 426 expected to be the most important variables affecting tree growth in the 427 Mediterranean region, characterized by hot and dry summers (Granda et al., 2013), 428 as water availability is considered to be the main limiting factor for plant growth 429 (Cherubini et al., 2003; Martínez-Vilalta et al., 2008). In fact, cambial activity is 430 usually affected (i.e., reduced, or stopped) during the summer season in these 431 regions (Cherubini et al., 2003). Nevertheless, Holm oak trees are able to maintain 432 carbon gain during this period of the year (Corcuera et al., 2005). The capacity of 433 Holm oak to tolerate the dry summer season is supported by our results that showed 434 significant positive responses to late spring – early summer precipitation and SPEI, 435 in accordance with previous studies (Corcuera et al., 2004; Gea-Izquierdo et al., 436 2011; Barbeta et al., 2013). 437 The climatic conditions during the winter period (i.e., usually cold) play an important 438 role in Mediterranean regions, highlighting the influence of the timing of precipitation 439 and temperature increases several months before growth resumption (Corcuera et 440 al., 2004; Natalini et al., 2016). Winter climatic variables had a positive effect on the 441 growth of all Holm oak trees, similar to the results reported by Granda et al. (2013). 442 These results point out that this species largely depends on favourable winter 443 climate conditions (Barbeta et al., 2013; Granda et al., 2013). For instance, enough 444 precipitation assures the crucial replenishment of soil water reservoirs (Campelo et 445 al., 2009; Gea-Izquierdo et al., 2011; Abrantes et al., 2013), and mild winter 446 temperatures may allow for extended periods of photosynthesis (Granda et al., 447 19 2014). Winter photosynthesis and the associated carbon gain allow Holm oaks to 448 recover carbon reserves (Gea-Izquierdo et al., 2011) and ensures the annual carbon 449 balance needed to form wood during the following growing season (Savé et al., 450 1999), and to minimize drought-induced decline (Galiano et al., 2012; Rosas et al., 451 2013). 452 The severe drought events registered in the study area (1983, 1986, 1989, 1992, 453 1995, and 2005) acutely affected tree growth independently of their defoliation state. 454 All Holm oak trees registered very low growth rates under these water stress 455 conditions, a typical pattern for oaks experiencing drought (Corcuera et al., 2004; Di 456 Filippo et al., 2010; Barbeta et al., 2013; Natalini et al., 2016). However, only two 457 years after the severe drought events ceased, Holm oaks, independent of their 458 crown condition, were able to recover to previous to drought growth rates, a pattern 459 previously found for this species (Granda et al., 2013). Drought-induced defoliation 460 rates in Quercus ilex are associated with low concentrations of stored non-structural 461 carbohydrates that may impede trees to recover after drought events (Galiano et al., 462 2012; Rosas et al., 2013). We have no proof of such depletion in our study 463 individuals, but given the resilience shown by all vigour groups, we suspect that 464 Holm oaks studied here either do not have this limitation or are able to rapidly 465 accumulate, mobilize and invert available carbohydrate reserves into growth when 466 conditions become favourable (Sala et al., 2012). The observed post-drought 467 rebound in Holm oak growth highlights the ability of this species to overcome 468 stressful conditions (i.e., drought) independently of its growth and current defoliation 469 rates. Nevertheless, despite the resilience shown by all vigour groups, the most 470 defoliated trees showed persistent low growth. In fact, the growth rates registered by 471 the dying individuals were so low that, despite recovery after the severe drought 472 20 events, these changes were not detected through the Changepoint analyses. Hence, 473 our results suggest that if drought events become more frequent and intense (IPCC, 474 2014), the time period needed for Holm oak reserves to recover may be too long, 475 leading to a progressive loss of resilience (Galiano et al., 2012) and increased 476 decline and/or mortality rates (Gea-Izquierdo et al., 2011; Granda et al., 2013; 477 Camarero et al., 2015b; Camarero et al., 2016; Natalini et al., 2016). 478 To conclude, our study shows that less vigorous Holm oaks (i.e., highly defoliated 479 and dying) grow significantly less than their more vigorous neighbours (i.e., healthy 480 and low defoliated). These patterns were persistent over the past three decades 481 (1980-2009) and suggest permanent growth reduction for the less vigorous 482 individuals. Reduced growth rates are frequently associated with the process of tree 483 mortality (Cailleret et al., 2017). Therefore, our results support the use of growth data 484 to detect early-warning signals of forest decline and swiftly design management 485 plans to mitigate climate change effects. Despite these differences, all vigour groups 486 responded similarly to climatic variability, with winter season having a strong effect 487 on Holm oaks growing in this Mediterranean coppice stand, and contributing together 488 with the spring and summer periods to the annual carbon budget and wood 489 formation during subsequent seasons. Moreover, the high resilience that all Holm 490 oak vigour groups have shown following severe drought events points out the 491 efficiency of this evergreen species to overcome stressful conditions. Still, as 492 reduced growth rates are frequently associated with the process of tree mortality, we 493 think that the less vigorous Holm oaks might not be able to cope with future water 494 stress conditions, leading to increased mortality rates among this emblematic 495 Mediterranean species. 496 497 21 Acknowledgements 498 The authors thank the coppice stand owners from Chapinería and Navas del Rey for 499 the permission of working on their property. We are particularly grateful to Dr. 500 Guillermo Gea-Izquierdo from INIA-CIFOR (Madrid, Spain) for allowing us to have 501 access to his Quercus ilex chronology (Gea-Izquierdo et al. 2009), which allowed us 502 to verify the consistency of our Holm oak tree-ring database. This work was 503 supported by the Spanish Ministry of Economy and Competitiveness (MINECO) with 504 the project VERONICA (CGL2013-42271-P), and by the Spanish Ministry for 505 Innovation and Science with the grant Consolider-Montes (CSD2008_00040). 506 507 REFERENCES 508 1. Abrantes, J., Campelo, F., García-González, I., Nabais, C., 2013. Environmental 509 control of vessel traits in Quercus ilex under Mediterranean climate: relating xylem 510 anatomy to function. Trees Struct Funct 27, 655-662, doi: 10.1007/s00468-012-511 0820-6. 512 2. Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., 513 Vennetier, M., Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H., Gonzalez, P., 514 Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.H., Allard, G., Running, 515 S.W., Semerci, A., Cobb, N., 2010. A global overview of drought and heat-induced 516 tree mortality reveals emerging climate change risks for forests. For Ecol Manage 517 259, 660-684, http://dx.doi.org/10.1016/j.foreco.2009.09.001. 518 3. Anderegg, W.R.L., Kane, J.M., Anderegg, L.D.L., 2013a. Consequences of 519 widespread tree mortality triggered by drought and temperature stress. Nat Clim 520 Change 3, 30-36, doi: 10.1038/nclimate1635. 521 22 4. Biondi, F., Waikul, K., 2004. A C++ program for statistical calibration of climate 522 signals in tree-rings chronologies. Computers & Geosciences 30, 303-311, 523 https://doi.org/10.1016/j.cageo.2003.11.004. 524 5. Biondi, F., Qeadan, F., 2008. A theory-driven approach to tree-ring 525 standardization: defining the biological trend from expected basal area increment. 526 Tree-Ring Research 64, 81-96, http://dx.doi.org/10.3959/2008-6.1. 527 6. Blanco, E., Casado, M.A., Costa, M., Escribano, R., García, M., Génova, M., 528 Gómez, A., Gómez, F., Moreno, J.C., Morla, C., Regato, P., Sáinz, H., 1997. Los 529 bosques ibéricos: una interpretación geobotánica. Planeta, Madrid 530 7. Brasier, C.M., 1996. Phytophthora cinnamomi and oak decline in southern 531 Europe. Environmental constraints including climate change. Ann For Sci 53, 347-532 358, doi: 10.1051/forest:19960217. 533 8. Bunn, A.G., 2008. A dendrochronology program library in R (dplR). 534 Dendrochronologia 26, 115-124, doi: http://dx.doi.org/10.1016/j.dendro.2008.01.002. 535 Bunn, A.G., Korpela, M., Biondi, F., Campelo, F., Mérian, P., Qeadan, F., Zhang, C., 536 2016. dplR: Dendrochronology Program Library in R. R package version 1.6.2. 537 9. Cailleret, M., Jansen, S., Robert, E.M.R., Desoto, L., Aakala, T., Antos, J.A., 538 Beikircher, B., Bigler, C., Bugmann, H., Caccianiga, M., Čada, V., Camarero, J.J., 539 Cherubini, P., Cochard, H., Coyea, M.R., Čufar, K., Das, A.J., Davi, H., Delzon, S., 540 Dorman, M., Gea-Izquierdo, G., Gillner, S., Haavik, L.J., Hartmann, H., Hereş, A.M., 541 Hultine, K.R., Janda, P., Kane, J.M., Kharuk, V.I., Kitzberger, T., Klein, T., Kramer, 542 K., Lens, F., Levanic, T., Linares Calderon, J.C., Lloret, F., Lobo-Do-Vale, R., 543 Lombardi, F., López Rodríguez, R., Mäkinen, H., Mayr, S., Mészáros, I., Metsaranta, 544 J.M., Minunno, F., Oberhuber, W., Papadopoulos, A., Peltoniemi, M., Petritan, A.M., 545 Rohner, B., Sanguesa-Barreda, G., Sarris, D., Smith, J.M., Stan, A.B., Sterck, F., 546 23 Stojanovic, D.B., Suarez, M.L., Svoboda, M., Tognetti, R., Torres-Ruiz, J.M., 547 Trotsiuk, V., Villalba, R., Vodde, F., Westwood, A.R., Wyckoff, P.H., Zafirov, N., 548 Martínez-Vilalta, J., 2017. A synthesis of radial growth patterns preceding tree 549 mortality. Glob Change Biol 23, 1675-1690, doi: 10.1111/gcb.13535. 550 10. Camarero, J.J., Gazol, A., Sangüesa-Barreda, G., Oliva, J., Vicente-Serrano, 551 S.M., 2015a. To die or not to die: early warnings of tree dieback in response to a 552 severe drought. J Ecol 103, 44-57, doi: 10.1111/1365-2745.12295. 553 11. Camarero, J.J., Franquesa, M., Sangüesa-Barreda, G., 2015b. Timing of drought 554 triggers distinct drought responses in Holm oak: implications to predict warming-555 induced forest defoliation and growth decline. Forests 6, 1576-1597, doi: 556 10.3390/f6051576. 557 12. Camarero, J.J., Sangüesa-Barreda, G., Vergarechea, M., 2016. Prior height, 558 growth, and wood anatomy differently predispose to drought-induced dieback in two 559 Mediterranean oak species. Ann For Sci 73, 341-351, doi: 10.1007/s13595-015-560 0523-4. 561 13. Campelo, F., Nabais, C., García-González, I., Cherubini, P., Gutiérrez, E., 562 Freitas, H., 2009. Dendrochronology of Quercus ilex L. and its potential use for 563 climate reconstruction in the Mediterranean region. Can J For Res 39, 2486-2493, 564 https://doi.org/10.1139/X09-163. 565 14. Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., Peñuelas, J., 2011. 566 Widespread crown condition decline, food web disruption, and amplified tree 567 mortality with increased climate change-type drought. PNAS 108, 1474-1478, doi: 568 10. 10.1073/pnas.1010070108. 569 15. Cherubini, P., Gartner, B.L., Tognetti, R., Bräker, O.U., Schoch, W., Innes, J.L., 570 2003. Identification, measurement and interpretation of tree rings in woody species 571 24 from Mediterranean climates. Biol Rev 78, 119-148, doi: 572 10.1017/S1464793102006000. 573 16. Cook, E.R., Peters, K., 1981. The smoothing spline: a new approach to 574 standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-575 Ring Bull 41, 45-53. 576 17. Cook, E.R., Holmes, R.L., 1986. Guide for computer program ARSTAN. 577 Laboratory of Tree-Ring Research, University of Arizona, Tucson. 578 18. Cook, E.R., Kairiukstis, L.A., 1990. Dendrochronology: applications in the 579 environmental sciences. Kluwer Academic Publishers, Doredrecht, the Netherlands. 580 19. Corcuera, L., Camarero, J.J., Gil-Pelegrín, E., 2004. Effects of a severe drought 581 on Quercus ilex radial growth and xylem anatomy. Trees Struct Funct 18, 83-92, doi: 582 10.1007/s00468-003-0284-9. 583 20. Corcuera, L., Morales, F., Abadía, A., Gil-Pelegrín, E., 2005. Seasonal changes 584 in photosynthesis and photoprotection in a Quercus ilex subsp. ballota woodland 585 located in its upper altitudinal extreme in the Iberian Peninsula. Tree Physiol 25, 599-586 608, doi: https://doi.org/10.1093/treephys/25.5.599. 587 21. Di Filippo, A., Alessandrini, A., Biondi, F., Blasi, S., Portoghesi, L., Piovesan, G., 588 2010. Climate change and oak growth decline: dendroecology and stand productivity 589 of a Turkey oak (Quercus cerris L.) old stored coppice in Central Italy. Ann For Sci 590 67, 706-706, doi: 10.1051/forest/2010031. 591 22. Dobbertin, M., 2005. Tree growth as indicator of tree vitality and of tree 592 reaction to environmental stress: a review. Eur J For Res 124, 319-333, doi: 593 10.1007/s10342-005-0085-3. 594 23. Fritts, H.C., 1976. Tree rings and climate. Academic Press, London, United 595 Kingdom. 596 25 24. Galiano, L., Martínez-Vilalta, J., Sabaté, S., Lloret, F., 2012. Determinants of 597 drought effects on crown condition and their relationship with depletion of carbon 598 reserves in a Mediterranean holm oak forest. Tree Physiol 32, 478-489, doi: 599 10.1093/treephys/tps025. 600 25. Gea-Izquierdo, G., Martín-Benito, D., Cherubini, P., Cañellas, I., 2009. Climate-601 growth variability in Quercus ilex L. west Iberian open woodlands of different stand 602 density. Ann For Sci 66, 802-802, doi: 10.1051/forest/2009080. 603 26. Gea-Izquierdo, G., Cherubini, P., Cañellas, I., 2011. Tree-rings reflect the impact 604 of climate change on Quercus ilex L. along a temperature gradient in Spain over the 605 last 100 years. For Ecol Manage 262, 1807-1816, doi: 606 http://dx.doi.org/10.1016/j.foreco.2011.07.025. 607 27. Gea-Izquierdo, G., Guibal, F., Joffre, R., Ourcival, J.M., Simioni, G., Guiot, J., 608 2015. Modelling the climatic drivers determining photosynthesis and carbon 609 allocation in evergreen Mediterranean forests using multiproxy long time series. 610 Biogeosciences 12, 3695-3712, doi: 10.5194/bg-12-3695-2015. 611 28. Giorgi, F., Lionello, P., 2008. Climate change projections for the Mediterranean 612 region. Glob Planet Change 63, 90-104, doi: 613 http://dx.doi.org/10.1016/j.gloplacha.2007.09.005. 614 29. Granda, E., Camarero, J.J., Gimeno, T.E., Martínez-Fernández, J., Valladares, 615 F., 2013. Intensity and timing of warming and drought differently affect growth 616 patterns of co-occurring Mediterranean tree species. Eur J For Res 132, 469-480, 617 doi: 10.1007/s10342-013-0687-0. 618 30. Granda, E., Scoffoni, C., Rubio-Casal, A.E., Sack, L., Valladares, F., 2014. Leaf 619 and stem physiological responses to summer and winter extremes of woody species 620 across temperate climate. Oikos 123, 1281-1290, doi: 10.1111/oik.01526. 621 32 Table 2 – Summary of the linear mixed-effects model (estimates ± SE) in which 746 logBAI varied as a function of the vigour group (highly, low defoliated, highly 747 defoliated, and dying), climatic variables (TJul(t-1)Aug(t), and PJul(t-1)Aug(t)), and the 748 interactions vigour group x climatic variables, and TJul(t-1)Aug(t) x PJul(t-1)Aug(t). 749 Model Estimates ± SE Intercept - 3.845 ± 1.69* Low defoliated 0.437 ± 1.23 Highly defoliated -1.295 ± 1.26 Dying 0.183 ± 1.31 TJul(t-1)Aug(t) 0.275 ± 0.10** Low defoliated x TJul(t-1)Aug(t) -0.017 ± 0.08 Highly defoliated x TJul(t-1)Aug(t) 0.063 ± 0.08 Dying x TJul(t-1)Aug(t) -0.020 ± 0.08 PJul(t-1)Aug(t) 0.020 ± 0.00*** Low defoliated x PJul(t-1)Aug(t) -0.000 ± 0.00 Highly defoliated x PJul(t-1)Aug(t) -0.000 ± 0.00 Dying x PJul(t-1)Aug(t) -0.000 ± 0.00 TJul(t-1)Aug(t) x PJul(t-1)Aug(t) - 0.001 ± 0.00*** Significant relationships at 0.05, 0.01, and 0.001 probability levels are market with *, 750 **, ***, and in bold. Abbreviations: SE, standard error; (t-1), previous to growth year; 751 (t), current year of growth. 752 753 33 Table 3 – Summary of the linear mixed-effects models (estimates ± SE) in which logBAI varied as a function of the vigour group (highly, low 754 defoliated, highly defoliated, and dying), climatic variables (SPEIAug(t-1)Jun(t), TNov(t-1)Feb(t), PNov(t-1)Feb(t), SPEINov(t-1)Feb(t), TMay(t)Jun(t), PMay(t)Jun(t), and 755 SPEIMay(t)Jun(t)), and the interactions vigour group x climatic variables. 756 Model (Climatic variable) Intercept Low defoliated Highly defoliated Dying Climatic variable Low defoliated x Climatic variable Highly defoliated x Climatic variable Dying x Climatic variable SPEIAug(t-1)Jun(t) 1.889 ± 0.13*** 0.021 ± 0.16 - 0.411 ± 0.17* -0.318 ± 0.17 0.803 ± 0.06*** -0.137 ± 0.08 -0.145 ± 0.08 -0.114 ± 0.09 TNov(t-1)Feb(t) 0.009 ± 0.25 0.161 ± 0.33 -0.439 ± 0.33 0.138 ± 0.35 0.244 ± 0.03** * -0.017 ± 0.04 0.005 ± 0.04 -0.060 ± 0.04 PNov(t-1)Feb(t) 1.471 ± 0.1 3 *** 0.115 ± 0.17 - 0.391 ± 0.18* -0.262 ± 0.19 0.002 ± 0.00*** -0.001 ± 0.00 -0.000 ± 0.00 -0.000 ± 0.00 SPEINov(t-1)Feb(t) 1.835 ± 0.13*** 0.028 ± 0.16 - 0.400 ± 0.17* -0.308 ± 0.17 - 0.190 ± 0.04*** 0.007 ± 0.05 0.024 ± 0.05 0.037 ± 0.05 TMay(t)Jun(t) 4.885 ± 0.35*** -0.021 ± 0.46 -0.683 ± 0.47 -0.622 ± 0.49 - 0.158 ± 0.02** * 0.003 ± 0.02 0.015 ± 0.02 0.016 ± 0.02 PMay(t)Jun(t) 1.578 ± 0.13*** 0.031 ± 0.17 - 0.377 ± 0.18* -0.246 ± 0.18 0.004 ± 0.00** * -0.000 ± 0.00 -0.000 ± 0.00 -0.001 ± 0.00 SPEIMay(t)Jun(t) 1.851 ± 0.13*** 0.029 ± 0.16 - 0.403 ± 0.17* -0.313 ± 0.17 0.253 ± 0.03** * -0.014 ± 0.03 -0.038 ± 0.03 -0.061 ± 0.04 Significant relationships at 0.05, 0.01, and 0.001 probability levels are market with *, **, ***, and in bold. Abbreviations: SE, standard error; (t-757 1), previous to growth year; (t), current year of growth. 758 34 FIGURE CAPTIONS 759 Fig. 1 – Holm oak defoliation (%) during the year of sampling (October 2010; marked with 760 a solid black arrow), and in the following two years (May 2011, April 2012, and October 761 2012) (left panel). Temporal (1980 – 2009) basal area increment (BAI) trends of healthy, 762 low defoliated, highly defoliated, and dying Holm oak trees, and of the SPEIAug(t-1)Jun(t) 763 (standardized precipitation-evapotranspiration index) (right panel); error bars show 764 standard errors. 765 Fig. 2 – Climate-growth relationships of healthy, low defoliated, highly defoliated, and 766 dying vigour groups, based on the correlation functions implemented in the DendroClim 767 2002 software (Biondi and Waikul 2004). Monthly climatic data are represented by 768 temperature (T°C; a), precipitation (P mm; b), and standardized precipitation-769 evapotranspiration index (SPEI; c). Time interval covers months from previous to growth 770 year (January [t-1] to December [t-1]; left panels) and current year of growth (January [t] to 771 December [t]; right panels). The vertical solid line separates between previous to growth 772 year (t-1) and current year of growth (t) periods. All given relationships are significant (p < 773 0.05). 774 Fig. 3 – Superposed epoch analyses showing growth (RWI, ring width indices) departures 775 from mean values given a set of key (severe drought events; year 0 on the x axis), and 776 lagged events: two years before (-2, and -1 on the x axis) and two after the severe drought 777 events (2, and 1 on the x axis). Values were calculated considering the severe drought 778 events, i.e. 1983, 1986, 1989, 1992, 1995, and 2005 (see Materials and Methods), 779 registered within the 1980 – 2009 study period. Black and grey columns indicate significant 780 (p < 0.05) and no significant (p > 0.05), respectively, growth departures from year 0, 781 considering random simulations. 782 Supplementary Fig. 1 – Quercus ilex L. subsp. ballota [Desf.] (Holm oak) wood core 783 showing this species’ diffuse-to semi-ring-porous wood with multiseriate rays, and tree-784 rings. 785 35 Supplementary Fig. 2 – Gaussian diagram (1973 – 2009) (a), and temporal (1980 – 786 2009) trends of temperature (T, °C) and precipitation (P, mm) (b). All climatic data were 787 available from the Cuatro Vientos meteorological station. 788 Supplementary Fig. 3 – Basal area increment (BAI) changes in mean values for healthy, 789 low defoliated, highly defoliated, and dying Holm oak trees, according to Changepoint 790 analyses results for the 1980 – 2009 period. Horizontal solid black lines mark the fitted 791 mean as it follows: if broken, then significant changes in mean are present (e.g., healthy, 792 low defoliated, and highly defoliated vigour groups); if continuous, then no significant 793 changes in mean are present (e.g., dying vigour group). 794