1 Sequence of plant responses to droughts of different time scales: lessons from holm 2 oak (Quercus ilex) forests 3 Adrià Barbeta1, 2*, Josep Peñuelas1, 2 1 4 CSIC, Global Ecology Unit CREAF-CSIC-UAB, E-08193 Cerdanyola del Vallès 5 (Catalonia), Spain; 2 CREAF, E-08193 Cerdanyola del Vallès (Catalonia), Spain. 6 *Corresponding author. Email: a.barbeta@creaf.uab.cat 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 Post-print of: Barbeta, A. and Peñuelas, J. “Sequence of plant responses to droughts of different timescales: lessons from holm oak (Quercus ilex) forests” in Plant ecology and 27 diversity (Ed. Taylor and Francis), vol. 9 issue 4 (2016) p. 321-338. The final version is available at DOI 10.1080/17550874.2016.1212288 28 Abstract 29 The functional traits of plants in regions of the world with a Mediterranean climate have been 30 shaped to tolerate periods of water deficit. These species are adapted to summer droughts but 31 may not be able to cope with future increases in drought intensity, duration, and/or 32 frequency. Here we review the mechanisms and traits of drought resistance and recovery of 33 the well-studied holm oak (Quercus ilex), which we propose as a model species for 34 Mediterranean-type ecosystems. Our aim was to understand the differences and links 35 between the responses of Q. ilex to summer droughts, extreme droughts, and long-term 36 drought experiments. A main goal was to provide an integral picture of drought responses 37 across organisational and temporal scales for identifying the most relevant processes that are 38 likely to contribute to determining the future of Mediterranean vegetation. Evidence from 39 long-term drought experiments showed that acclimation processes from the molecular (e.g. 40 epigenetic changes) to the ecosystem level (e.g. reductions in stand density) mitigate the 41 effects of drought. Changes in leaf morphology and hydraulics, leaf-to-shoot allometry and 42 root functioning are among the key mechanisms for overcoming increasing drought. The 43 duration of drought determines its severity in terms of canopy loss and stem mortality. 44 Although Q. ilex can vigorously resprout after such episodes, its resilience may be 45 subsequently reduced. In the future, higher frequency of return of extreme droughts will 46 challenge thus the capacity of these forests to recover. The insights provided by this review of 47 the complex interplay of processes that determine the response of trees to droughts of 48 different duration, intensity, and frequency will also help us to understand the likely responses 49 of other resprouting angiosperms in seasonally dry ecosystems that share similar functional 50 traits with Q. ilex. 51 Keywords: acclimation, climate change, defoliation, drought stress, Quercus ilex, temporal 52 scales, water availability. 53 54 Introduction 55 The increasing length and intensity of droughts is producing dieback in all forested regions of 56 the world (Allen et al. 2010, 2015). The vulnerability of forests to droughts of a certain length 57 and intensity depends largely on the climate of the region to which the structure and function 58 of the forests are adapted (Vicente-Serrano et al. 2013). Tree species have evolved to cope 59 with climatic variability (including the variability of drought duration), so the mechanisms to 60 resist drought have been adjusted to the climatic variability during the evolutionary histories 61 of the trees. Longer, more intense droughts are forcing trees to display their phenotypic 62 plasticity to its maximum (Matesanz and Valladares 2013). Severe modifications of tree 63 function or forest structure may often occur, however, when drought reaches a certain 64 threshold of length and/or intensity. Furthermore, the capacity of recovery after such a 65 disturbance may also be influenced by prior disturbances and by the length of the period until 66 a potential subsequent disturbance (Anderegg et al. 2015). The mechanisms of drought 67 resistance, the responses to severe and extreme droughts, and the subsequent recovery 68 include processes from the molecular to the stand level, which operate at different timescales. 69 A focus on the temporal dimension of both the abiotic (duration and intensity of droughts) and 70 biotic (drought-resistance mechanisms and drought effects on plants) factors would thus 71 improve our ability to predict potential changes in the structure and function of vegetation. 72 In this review, we discuss the effect of drought on a tree species, from the tissue level 73 to stand or ecosystem or level, at range of time scales. Our chosen model tree is the holm oak 74 (Quercus ilex L.), an extensively studied Mediterranean forest tree. Q. ilex possesses some 75 adaptations to drought which the species shares with many other species of Mediterranean- 76 type ecosystems. It is a keystone species in vast areas of the western Mediterranean Basin, 77 and its dominance is expected to increase to the detriment of other forest species that appear 78 to be more sensitive to aridity, and in response to land abandonment, followed by secondary 79 regrowth by Q. ilex (Carnicer et al. 2014). The numerous studies on the responses of Q. ilex to 80 drought have allowed us to construct a case study to understand the drought-induced 81 physiological and ecological processes that all vary in their response times. We explore 82 changes in forest structure and function, including the different feedbacks that result from 83 extreme drought episodes and a sustained increase in aridity. 84 The World Meteorological Organisation (1992) provides two different definitions of 85 drought. The first one defines drought as a prolonged absence or marked deficiency of 86 precipitation. In the regions with a Mediterranean climate, the period with the lowest 87 precipitation coincides with the warmest temperatures every year. This is referred to summer 88 drought in this review, and implies weather conditions similar to the climatic average for a 89 given site. However, drought is also defined as a period of abnormally dry weather, sufficiently 90 long to cause a serious hydrological imbalance. We call this second definition an extreme 91 drought in this study. We also review the response of holm oak forests to drought periods in 92 which the aridity is created or exacerbated by abnormally warm weather (Allen et al. 2015). 93 We first discuss the resistance mechanisms that operate during the summer droughts 94 that typically occur in Mediterranean regions and which last three to four months. Second, we 95 review the effects of droughts that exceed the average climatic variability of a given location in 96 terms of duration, seasonality, intensity, or coincide with heat waves. Third, we report the 97 effects of persistent droughts, such as those produced by experimental rain exclusions. We 98 demonstrate how this information is useful for a better understanding of the response of Q. 99 ilex to environmental change, but also to show how the temporal scale is crucial for predicting 100 the responses of plants to the increases in drought projected by global circulation models 101 (IPCC 2013). The mechanisms and traits that Q. ilex possesses to resist and recover from 102 droughts (Table 1) are shared by many plant species from Mediterranean-type ecosystems and 103 other drought-prone ecosystems, especially by resprouting angiosperms (Zeppel et al. 2015). 104 Therefore, the insights provided by this review are, to some extent, representative of the 105 ecology of Mediterranean vegetation under current climate and can serve for indicating 106 expected climate change responses. 107 1.1 Quercus ilex as a paradigmatic Mediterranean species. 108 The insights from published studies provided by the present synthesis on the effect of drought 109 at different timescales may be relevant not only for Q. ilex conservation and management but 110 also for woody Mediterranean flora, especially resprouting angiosperms. Q. ilex shares many 111 morphological, physiological, and phenological traits related to the response to water stress 112 (Table 1) with other species from regions with a Mediterranean climate, such as in California, 113 Chile, Australia, or South Africa. The maximum rooting depth of holm oak is similar to the 114 global average of species in sclerophyllous forests and shrublands (Canadell et al. 1996) 115 whereby high investments in below-ground biomass are common in these seasonally dry 116 ecosystems. Similarly, sclerophyllous leaves are also a typical feature of Mediterranean 117 vegetation, indicating the convergent evolution of plant form by several selective forces, 118 including drought (Mooney and Dunn 1970). Mediterranean summer drought stops growth in 119 Q. ilex, resulting in bimodal secondary growth patterns (Lempereur et al. 2015). A transient 120 quiescence of growth has also been reported for Mediterranean conifers (Camarero et al. 121 2010; Vieira et al. 2014) and forbs (Olano et al. 2013). Large below- and above-ground bud 122 banks are also characteristically associated with Mediterranean vegetation (Canadell and 123 López-Soria 1998; Klimešová and Klimeš 2007) and its capacity to recover from fires but are 124 also relevant after extreme droughts (Zeppel et al. 2015). Furthermore, resprouting species 125 share a suite of ecophysiological and anatomical traits (Hernández et al. 2011, Vilagrosa et al. 126 2014), so their response to drought should be similar, albeit with differences in sensitivity. 127 Mediterranean summer droughts and the adaptive mechanisms and traits of Q. ilex 128 Plants in Mediterranean ecosystems have to cope with a double limitation or stressful period: 129 summer drought and winter cold (Terradas 1999). The high summer temperatures increase 130 evapotranspiration, which, together with the low precipitation, produce periods with a 131 negative water balance (the difference between precipitation and water losses through 132 evapotranspiration and runoff). The summer drought, within the distributional range of forests 133 dominated by Q. ilex, typically lasts three to four months (Figure 1). The duration and intensity 134 of the summer droughts, however, are highly variable between years (Piñol et al. 1991), and 135 some Q. ilex forests thrive in circum-Mediterranean areas where summer drought is rare, such 136 as the coastal areas of the Bay of Biscay (Niinemets 2015), and evidence suggests that 137 colonisation in areas with an oceanic climate is recent (Delzon et al. 2013)(Figure 2). Pollen 138 records indicate that Q. ilex persisted during the last glacial period in climatic refugia in the 139 wettest part of its current distribution and gained early dominance in the Holocene in more 140 semi-arid areas of current Morocco and southern Spain (Reille and Pons 1992). Human activity 141 may have contributed to the expansion of Q. ilex in the wettest areas of its distribution during 142 the second half of the Holocene (Reille and Pons 1992). Furthermore, this tree species will 143 likely expand its distribution in the future as the climate becomes warmer and the availability 144 of water decreases. 145 Q. ilex possesses important ecotypic variation (Gratani et al. 2003, Peguero-Pina et al. 146 2014) but is generally considered as a species that has several morphological and physiological 147 adaptations for surviving summer droughts (Table 1). Summer drought depletes the water in 148 the topsoil, but moisture can still be found in deeper soil layers. Because water is taken up by 149 roots, the development of a deep root system in Q. ilex has been regarded as a morphological 150 adaptation to summer drought (Canadell et al. 1999). Woody species in sclerophyllic forests 151 and shrublands (Canadell et al. 1996) also have deeper roots than plants in ecosystems 152 without seasonal drought. The effect of drought at sites with shallow soils is determined by the 153 capacity of roots to exploit fissures in the bedrock to access moisture pockets (Lloret et al. 154 2004; Barbeta et al. 2015). Deep roots are thus a crucial morphological feature of holm oaks, 155 conferring a competitive advantage over sympatric species, such as Mediterranean pines 156 (Ferrio et al. 2003, Comas et al. 2015) . This ability to tap deep water tables and thus have 157 permanent access to water has also been confirmed at the ecosystem level by data for 158 synchronous eddy covariance and soil water content (Reichstein et al. 2002). The possession of 159 this morphological character has led to the categorisation of Q. ilex as a drought-avoiding 160 species (David et al. 2004). Tapping deep water table is maximised where soils lay on 161 permeable or fractured bedrock in a relatively flat terrain and with sufficient groundwater 162 recharge during winter months (David et al. 2004). In addition to the depth of the root system, 163 the ratio of total root biomass relative to above-ground biomass is associated with water 164 availability. Holm oaks in xeric sites have a higher percentage of below-ground biomass than 165 those in mesic sites; roots account for more than half the total tree biomass in some cases 166 (Canadell et al. 1999). Fine roots have been shown to have a higher relative area in the drier 167 sites of Q. ilex (Martin-Stpaul et al. 2013). The variability in the length of summer drought is 168 thus offset by the allocation of resources for the growth of below-ground structures to the 169 detriment of aerial organs. As a result, Q. ilex trees withstand the typical summer droughts by 170 ensuring access to water through their extensive root systems and also by maintaining a 171 sufficiently low ratio of above-ground to below-ground biomass. 172 The morphology of aerial organs has been evolutionarily shaped by summer drought to 173 promote a conservative use of water. Xylem conduits in Q. ilex are narrower than in Quercus 174 deciduous tree species (Villar-Salvador et al. 1997), resulting in lower conductivity, although 175 this may be an adaptation to winter cold (Cavender-Bares et al. 2005). In contrast, Q. ilex have 176 wider xylem conduits compared to sympatric Mediterranean evergreen species from other 177 genus (Martínez-Vilalta et al. 2002), and thus transpire at higher rates than those. Sap flow 178 during summer drought, however, must be substantially reduced (Martínez-Vilalta et al. 2003; 179 Barbeta et al. 2012) to avoid xylem cavitation, because the xylem anatomy of Q. ilex is not 180 more resistant to embolism than that of deciduous Quercus species in temperate forests 181 (Choat et al. 2012). Once water reaches the canopy, sclerophyllous leaves with high specific 182 mass and thickness promote the conservation of cellular water (Bartlett et al. 2012). Together, 183 the morphologies of roots, xylems, and leaves partly determine the water-use strategy of Q. 184 ilex and thus its performance under drought. 185 When morphological adaptations to dry environments are not sufficient to avoid 186 completely any physiological water stress, plants activate several physiological mechanisms, 187 most relevantly stomatal control (Martínez-Vilalta et al. 2014). Q. ilex may be the most 188 isohydric species of the genus, and more isohydric than sympatric sclerophyllous shrubs, 189 according to a recent review (Martínez-Vilalta et al. 2014). Accordingly, it will reduce stomatal 190 conductance to avoid conductivity losses at less negative water potentials than other 191 congeneric or co-occuring species. Indeed, Q. ilex uses water very conservatively during 192 summer drought, with almost completely closed stomata at midday (Ogaya, Llusià, et al. 193 2014), and the trees consequently have a lower risk of xylem embolism. Nevertheless, 194 embolism in Q. ilex xylem is not rare and can occur either during summer drought (Aguadé et 195 al. 2015) or winter frost (Cavender-Bares et al. 2005). The soluble fraction of non-structural 196 carbohydrates has been hypothesised to be used for maintaining the integrity of the vascular 197 system by osmotic regulation (Sala et al. 2012). While it has not been observed in Q. ilex, the 198 unloading of sugars from the phloem to embolised xylem vessels has been proposed as a 199 possible mechanism of embolism repair (Nardini et al. 2011). Xylem refilling in Mediterranean 200 plants such as Olea europaea and Laurus nobilis has recently been shown to be ion-mediated 201 (Trifilò et al. 2014). Interestingly, the proportion of soluble sugars increases during summer 202 drought in the branches (Rosas et al. 2013) and leaves (Rivas-Ubach et al. 2014) of Q. ilex, 203 suggesting a role for these solutes in drought resistance. Rivas-Ubach et al. (2014) have also 204 found an increase in foliar potassium concentration in summer, which may also be related to 205 osmotic regulation (Babita et al. 2010). 206 Common with a number of other species, in Q. ilex thresholds of water availability (van 207 der Werf et al. 2007) and temperature (Deslauriers et al. 2008) constrain cambial activity and 208 thus govern the phenology of growth (Mitrakos 1980, Lempereur et al. 2015). The phenology 209 of primary, secondary, and reproductive growth are thus linked to summer drought and winter 210 cold. A bimodal pattern of stem growth that peaks in spring and autumn is common in Q. ilex 211 (Montserrat-Martí et al. 2009, Gutiérrez et al. 2011). The onset of the summer cessation of 212 growth has been correlated with spring rainfall (Lempereur et al. 2015). This quiescence might 213 be an adaptation to summer drought, but growth would be expected to be plastic enough to 214 continue when summers are sufficiently wet (Albuixech et al. 2012). Primary growth, including 215 foliar flushing (Ogaya and Peñuelas 2006) and shoot growth (Camarero et al. 2015), occurs 216 mostly before the summer drought, although foliar maturation and shoot growth are 217 completed in summer (Montserrat-Martí et al. 2009). Q. ilex thus takes advantage of periods 218 with favourable conditions for growth (Albuixech et al. 2012). In contrast to deciduous 219 Mediterranean oaks, it has a peak of growth in autumn, either of the stems (Corcuera et al. 220 2004, Gutiérrez et al. 2011, Albuixech et al. 2012) or the leaves (Ogaya and Peñuelas 2006). In 221 addition, several cohorts of buds up to 3-4 years old are maintained in the branches, which 222 may burst in years with favourable conditions (Alla et al. 2012). Overall, Q. ilex has large 223 phenological plasticity of growth-related traits to cope with the marked seasonality in water 224 supply of Mediterranean ecosystems. 225 The various morphological, physiological adaptations to drought and phenological 226 patterns in response to water shortage manifest themselves at a range of time scales (Table 1). 227 Above-ground to below-ground biomass ratios reflect the compound effects of environmental 228 conditions experienced during the lifetime of individuals. For example, contrasting micro- 229 topographical conditions can produce morphological differentiation between genetically 230 closely related populations in terms of allocation to roots and above-ground organs (Canadell 231 et al. 1999). The environmental conditions experienced during an individual's lifetime thus 232 contribute to defining its structure and functioning. However, foliar and xylem morphology are 233 determined mainly by the environmental conditions of the relevant periods of development - 234 spring for leaves (Ogaya and Peñuelas 2006) and summer for xylem (Abrantes et al. 2013). 235 Physiological mechanisms can be activated at even shorter times; the degradation of starch to 236 sugars occurs on a daily basis (Patakas et al. 2002), and stomatal response, mediated by … is 237 almost instantaneous (Mcadam and Brodribb 2012). Most of these mechanisms are transient 238 responses and can be reversed when the limitation disappears (sensu Wolkovich et al. (2014), 239 Table 1). For example, at the leaf-level, stomata can open fully again once soil moisture 240 availability increases, or leaves initiated in wet years may grow less sclerophyllic than those in 241 dry years. However, other features, such as modified xylem anatomy (e.g. the vessel area and 242 density) are persistent responses. Xylem anatomy fluctuates in response to the environmental 243 conditions during the growing season (Campelo et al. 2010), but it influences future water 244 transport because water is conducted by the xylem formed over several years (Martínez-Vilalta 245 et al. 2003). In those traits that form over longer periods, the plasticity to respond to sudden 246 climatic extremes is thus limited. 247 Response of Q. ilex to extreme drought 248 Summer droughts in Mediterranean ecosystems occur every year but with high inter-annual 249 variability in intensity and duration. However, as climatic change is increasing temperatures 250 and their variability (IPCC 2013), a positive trend effect (Jentsch et al. 2007) is likely to increase 251 the intensity of summer droughts. Extreme droughts, in addition to the cyclic summer 252 droughts, are also expected to become more frequent in the Mediterranean Basin (IPCC 2013; 253 Wolkovich et al. 2014). Droughts occurring in or extending into the typically wet seasons in the 254 Mediterranean Basin (spring and autumn) have recently increased and these extreme drought 255 events may have larger impacts than more intense summer droughts, because spring and 256 autumn are the optimum periods for growth in holm oaks (Camarero et al. 2015). 257 Extreme droughts are episodes that (i) represent statistically rare events regarding 258 their intensity, duration, and/or timing and that (ii) cause negative effects that involve at least 259 partial crown damage or an increase in tree mortality rates above the baseline levels. The 260 available literature on drought-induced dieback of holm oaks has reported a great diversity of 261 drought characteristics. Several extreme droughts have been recorded recently in the Iberian 262 Peninsula. For example, the 1993-1995 drought in the north-eastern Iberian Peninsula lasted 263 for 10 months, during which time precipitation was only half of the historical average 264 (Peñuelas et al. 2000, Lloret et al. 2004). In 2005 drought in the southern Pyrenees was very 265 intense and aseasonal, peaking in the spring, while another in the same region in 2012 was 266 characterised by less intense but lasted longer (Camarero et al. 2015). Other, more local 267 droughts - such as the drought in southern Catalonia in 2011 - had severe effects on the 268 vegetation but were restricted to smaller geographic areas (Poyatos et al. 2013, Ogaya, 269 Barbeta, et al. 2014). All these events had in common the rarity of the historical climatic series, 270 either in drought duration, intensity, and/or timing. Q. ilex populations will suffer important 271 alterations in their use of water and carbon, phenology, primary and secondary growth rates, 272 or mortality rates, depending on the characteristics of the drought. 273 Drought-induced forest dieback of Q. ilex. 274 Leaf shedding occurs in Q. ilex simultaneously with bud burst (Ogaya and Peñuelas 2006), 275 whereby some of the old leaves are abscised and substituted by new leaves. Leaf shedding can 276 also be a short-term response to drought, whereby plants adjust their transpiratory area to 277 decreased water availability. Leaf abscission may be able to act as a safety valve to avoid 278 extensive embolism in the xylem of stems, where carbon investment is more expensive (Bucci 279 et al. 2012, Pivovaroff et al. 2014). Some level of leaf shedding during a summer drought is 280 thus not necessarily indicative of strong water stress and could be added to Table 1 as a 281 mechanism for regulating the plant-water relations. High foliage loss, however, reduces 282 photosynthetic capacity and may compromise the survival of trees in the long term (Galiano et 283 al. 2012). Extensive canopy loss is thus the first indicator of drought-induced forest dieback. 284 The crown defoliation of Q. ilex stands in the last 15 years has been correlated with a 285 progressive increase in water stress (Carnicer et al. 2011) and with extreme droughts 286 (Peñuelas et al. 2000, Lloret et al. 2004, Galiano et al. 2012, Ogaya, Barbeta, et al. 2014, 287 Barbeta et al. 2015). These events were noticeable by the browning of leaves and the 288 desiccation of entire branches, and, in some cases, of all above-ground organs of an individual. 289 Temperature has increased in most parts of the world over the last 100 years, but no such 290 consistent trend has been observed for precipitation (IPCC 2013). As a result, the vulnerability 291 of forests to drought is increasing because hotter temperatures enhance drought severity 292 (Allen et al. 2015). Therefore, the inclusion of temperature in the calculation of a drought 293 index is required for comparing the conditions of current and past droughts (Vicente-Serrano 294 et al. 2010). 295 The timing of crown defoliation or stem mortality triggered by drought has been 296 variable (Table 2); some events (1994 and 2011) were produced by long-lasting water deficits 297 (Lloret et al. 2004; Barbeta et al. 2015), whereas others caused a much more negative water 298 balance but for shorter periods (2005) (Barbeta et al. 2013). Stem mortality after the event 299 reported by Barbeta et al. (2013) was 5.4% (short-term acute water shortage) , but the 300 prolonged drought in 1994 caused 51.3% of stem mortality in stands over the same bedrock, 301 suggesting that Q. ilex has more tolerance to short-term and acute droughts. Confirming this 302 premise will require increasing observational efforts and understanding other factors inherent 303 to forest structure and function that interact with drought to determine the magnitude of its 304 effects. Furthermore, the lack of standardisation between the units to measure the effects of 305 drought on vegetation (crown defoliation at plot or individual level, or mortality rates at stem- 306 or individual-level) prevents a straightforward comparison among events. For example, 307 Barbeta et al. (2013) assessed mortality at the stem level as the lack of resprouting the year 308 after the drought event, because the isolation and identification of genets was not possible. In 309 contrast, Galiano et al. (2012) categorised trees (multi-stemmed individuals) into classes of 310 crown damage and assessed mortality at the genet level in another site. Harmonising the 311 metrics of the effects of drought on forests to provide comparable datasets would be desirable 312 in future studies and useful for assessing large-scale patterns. 313 Drought characteristics alone (represented in Table 2 by the multi-scalar drought 314 index) do not appear to fully explain the variability in crown defoliation or stem mortality. The 315 effects of a drought event depend on forest structure (Carnicer et al. 2011). Trees in denser 316 stands are more likely to die or be defoliated (Carnicer et al. 2011; Galiano et al. 2012). Having 317 more stems increases the survival of a multi-stemmed individual but increases individual stem 318 mortality and defoliation (Galiano et al. 2012). Taller trees (Lloret et al. 2004) with larger basal 319 areas (Galiano et al. 2012; Barbeta et al. 2013) are the least affected during these events. The 320 form of tree growth (multi-stems, in particular) is affected by management practices and thus 321 has an impact on drought survival. Topography is another critical factor that affects the levels 322 of mortality and defoliation in Q. ilex. The fluctuations of climate are offset, to some extent, by 323 the deep root system of Q. ilex (Canadell et al. 1999), but rooting depth can be spatially 324 heterogeneous. Soil is rockier, shallower, and has a lower capacity to store water on steep 325 slopes than on flat ground or in valley bottoms, so the survival of the trees varies across the 326 landscape (Galiano et al. 2012). Similarly, access to groundwater is critical for surviving 327 summer drought (Barbeta et al. 2015), and the mortality rates during an extreme drought are 328 substantially higher on bedrock types that do not allow the penetration of roots in fissures 329 (Lloret et al. 2004). 330 Trends of crown defoliation and their relationship with drought characteristics 331 To identify the more critical drought characteristics (in terms of duration and timing) for crown 332 defoliation (i.e. for Q. ilex resistance to drought), we carried out a correlative analysis, using 25 333 years of data on crown defoliation in the Iberian Peninsula (ICP (International Co-operative 334 Programme on Assessment and Monitoring of Air Pollution Effects on Forests), (Carnicer et al. 335 2011, de la Cruz et al. 2014)) and the global Standardised Precipitation Evapotranspiration 336 Index (SPEI) database (Vicente-Serrano et al. 2010). Using an updated version of the data 337 analysed by Carnicer et al. (2011) but only for Q. ilex, we correlated plot-level data on the 338 annual relative change of crown defoliation with gridded (0.25°) SPEI values at all different 339 months and temporal scales (periods ranging from 1 to 24 months). The correlation heat map 340 presented in Figure 3 indicates that the SPEI and the annual relative change in defoliation level 341 were best correlated for SPEI scales longer than seven months and during summer months 342 (June, July, and August). More particularly, June SPEI-23 (which is calculated using the water 343 balance, P-PET, of the 23 months prior to June) was best (negatively) correlated, followed by 344 July SPEI-21 and July SPEI-15. These data were collected from 25 years (1987-2013) and were 345 based on 233 plots distributed throughout the range of Q. ilex in the Iberian Peninsula. 346 The sensitivity of Q. ilex to drought indicated by crown defoliation was similar to the 347 sensitivity of secondary growth, indicated by tree-rings (Pasho et al. 2011, Camarero, 348 Sangüesa-Barreda, et al. 2015). The correlation between crown defoliation and the SPEI 349 became stronger by June and when accounting for the water balance of the previous winter 350 and spring (Figure 3). In those tree-ring studies (Gea-Izquierdo et al. 2009, Pasho et al. 2011, 351 Camarero, Sangüesa-Barreda, et al. 2015), growth rates were correlated highest with the 352 water balance in summer. In contrast, crown defoliation was better correlated with longer 353 periods of the water balance (i.e. timescales of the SPEI longer than seven months), spanning 354 nearly 2 years. This is in agreement with previously reported drivers of defoliation in Iberian 355 tree species at the large-scale, which found that the duration of drought is a statistically 356 significant factor (de la Cruz et al. 2014). The good correlations of defoliation with the SPEI 357 over long periods are likely to be a result of tapping of groundwater by Q. ilex (Reichstein et al. 358 2002, David et al. 2004, Barbeta et al. 2015). Certainly, the level of groundwater reserves 359 responds to the accumulated water balance of longer periods than the level of surface soil 360 moisture (Vicente-Serrano et al. 2010). Surface soil moisture is subject to strong short-term 361 fluctuations responding to single events of rainfall and to the atmospheric evaporative 362 demand. In contrast, deeper soil layers containing groundwater are recharged during wet and 363 cold periods and are less afected by evaporation. Growth is mainly dependent on the climate 364 of the growing climate that determines the surface soil moisture whereas, groundwater 365 reserves that have accumulated over periods of more than one year are crucial to survive 366 those droughts that deplete surface soil moisture completely. 367 Winter rainfall may also be related to the replenishment of tree carbon reserves. Sap 368 flow is higher in winter than in summer, but trees do not allocate carbon to above-ground 369 growth in winter (Sánchez-Costa et al. 2015). Thus carbon assimilated during winter is likely to 370 be directed to other functions, such as defence or cold hardening. Such winter carbon could 371 be also stored in the form of starch that may then be remobilised and used during summer 372 drought events to maintain cell turgor and the integrity of the vascular system (Table 1). 373 The sustained increase in crown defoliation in Q. ilex stands since the early 1990s 374 peaked during the droughts of 1994-1995 (Figure 4a). The defoliation levels have since 375 stabilised at around 20%. The SPEI values indicated that droughts of similar intensity occurred 376 after this peak (e.g. 2001 and 2005) but did not trigger larger increases in defoliation than 377 those in the mid-1990s, indicated by the relative change in crown defoliation (Figure 4b). 378 These data suggest that increasing aridity has led to a diminished canopy cover, but also that 379 once a certain level of defoliation was reached, droughts of equal intensity did not further 380 decrease leaf area. A reduction in leaf area to sapwood area and/or in the ratio of above- and 381 below-ground biomass prompted by droughts could be maintained in the long term. Forests 382 have been hypothesised to acclimate through stabilising processes at the individual (Barbeta et 383 al. 2015) and community levels (Lloret et al. 2012, Barbeta et al. 2013). Other studies, 384 however, have suggested that consecutive droughts can weaken trees and compromise their 385 survival in the long-term (Anderegg et al. 2015). The results of long-term drought experiments 386 (see below) and long-term studies on the recovery of affected stands following extreme 387 drought events can help to elucidate whether it is the case of stabilising processes or rapid 388 ecosystem change. In addition to the likely acclimation of forest leaf area as a result of 389 reduced water availability, a potential sampling-effect on the quantification of defoliation 390 should no be ruled out. Defoliation within the ICP forests programme (Carnicer et al. 2011, de 391 la Cruz et al. 2014) is estimated in relation to a reference tree supposed to be completely 392 healthy. Over the years, these reference trees may have also suffered some defoliation while 393 still being the relatively healthiest at the plot-level. This would have led to the 394 underestimation of defoliation in the most recent years. 395 Resprouting after drought-induced biomass loss and the legacy of droughts 396 Q. ilex has the capacity to resprout from pre-existing buds after a disturbance that causes the 397 loss of biomass (Espelta et al. 1999). Resprouts can emerge either from above-ground (canopy) 398 or below-ground (lignotuber) buds, allowing a rapid recovery of the photosynthetic area after 399 a disturbance. Fire, overgrazing, firewood extraction, and coppicing have been the most 400 common disturbances triggering resprouting in Q. ilex stands historically, and extreme 401 droughts will likely increasingly frequent in the coming years, and Q. ilex will likely resprout in 402 response. Extreme droughts cause low tree mortality compared to other disturbances such as 403 wildfires (Espelta et al. 1999). However, extreme droughts can impact roots by causing 404 hydraulic dysfunction (Anderegg et al. 2012, Meinzer and McCulloh 2013) and deplete below- 405 ground reserves of non-structural carbohydrates (NSC) (Galiano et al. 2012) and weaken the 406 resprouting capacity of trees. Still, Q. ilex has shown full canopy recovery within a year after an 407 extreme drought (Ogaya, Barbeta, et al. 2014) that induced extensive branch desiccation (Liu 408 et al. 2015). The presence of long-lived apical buds confers Q. ilex the capacity to resprout 409 from stems and twigs (Table 1, Alla et al. (2012)). The resilience of trees to drought relies on 410 these above-ground meristemic tissues that are not usually damaged by drought. 411 Despite the higher resistance to drought-induced mortality of resprouting plants 412 relative to non-resprouting plants (Zeppel et al. 2015), the resilience of the former can be 413 weakened if the frequency of extreme droughts increases. Lloret et al. (2004) reported that 414 plants that resprouted weakly after the 1985 drought were more likely to die in the drought of 415 1994. Forty percent of the plants that completely lost their green foliage in 1994 had not 416 recovered one year after the end of the drought. In contrast, only 11.2% of the trees that 417 maintained some of their leaves after the 1994 drought were completely defoliated the year 418 after (Lloret et al. 2004). Past droughts thus have a legacy, a carry-over effect, which has been 419 associated with the levels of NSC in lignotubers, the below-ground swollen structures that 420 accumulate starch and form buds (Canadell et al. 1999). Galiano et al. (2012) were the first to 421 demonstrate that the resilience of crown condition 7 years after an extreme drought was 422 associated with the amount of NSC in the lignotubers. NSC reserves were depleted 60% more 423 in highly damaged than in healthy trees. Rosas et al. (2013) observed significantly lower 424 lignotuber NSC levels in the most defoliated trees after the drought of 2011, confirming this 425 organ as an important NSC reservoir for overcoming the effects of drought. The period of 426 return of extreme drought appears to be a key factor for determining the resilience of Q. ilex 427 to these droughts. If lignotuber NSC levels are not fully recovered by the following drought, the 428 survival of the tree will be compromised. Q. ilex needs an estimated 20 years to fully replenish 429 the NSC levels of its lignotubers (López et al. 2009), so the current frequency of extreme 430 droughts (see time-lapse between droughts of similar intensity in Figure 2) is already 431 threatening the future resilience of Q. ilex stands. These extreme droughts will have tangible 432 effect on the levels of NSC reserves only if they result in extensive or total loss of the 433 photosynthetic area, impairing the use of newly assimilated carbon. It is known that the levels 434 of NSC reserves are not constrained by the source activity (i.e. the photosynthetic activity) and 435 that carbon acquisition can be maintained during periods of drought or low temperatures 436 (Körner 2003). Therefore, droughts that drastically reduce or stop growth but do not produce 437 leaf shedding will have little effect on the NSC reserves and consequently, the capacity of a 438 tree to resprout in response to drought will remain intact. 439 The hydraulic deterioration produced by drought-induced xylem cavitation 440 accumulates and persists over several years in aspen trees (Anderegg et al. 2013). This so- 441 called cavitation fatigue has been shown to increases the vulnerability of the xylem to 442 embolism of previously drought-affected stems in several species (e.g. Populus angustifolia, P. 443 tremuloides or Helianthus annuus), but in others (e.g. Acer negundo and Alnus incana) stems 444 do not suffer any change in vulnerability (Hacke et al. 2001). Other woody plants (Ligustrum 445 vulgare and Viburnum lantana) are even able to acclimate to drought by increasing their 446 resistance to embolism (Beikircher and Mayr 2009). This does not seem to be the case of 447 Quercus ilex, as was shown by a long-term drought experiment where its vulnerability of xylem 448 to cavitation has slightly but not significantly increased (Limousin, Longepierre, et al. 2010). 449 During extremely dry years, such as 1994 in the Iberian Peninsula, the vessels produced by Q. 450 ilex present a smaller diameter than during average years (Corcuera et al. 2004). Smaller 451 diameters imply a reduction in the hydraulic conductance and this persists for several years 452 (Table 1, (Martínez-Vilalta et al. 2003)). It is highly likely thus that Q. ilex trees will carry over a 453 drought legacy not only in their transpiring area and the levels of NSC but also in their 454 hydraulic function. This legacy will negatively predispose trees to maintain an adequate 455 hydraulic functioning by reducing the potential conductance and increasing the vulnerability to 456 embolism of xylem in the stems. 457 In addition to the drought legacies related to carbon and water use, drought may also 458 predispose trees to suffer attacks from insects and fungal pathogens (Desprez-Loustau et al. 459 2006, Oliva et al. 2014). Importantly, biotic agents are considered a contributing factor in 460 drought-induced forest dieback (Worrall et al. 2010). The combined effect of fungal infections 461 and drought-induced defoliation has been associated with lower growth, sapwood depth and 462 reduced conductivity in Pinus sylvestris individuals (Aguadé et al. 2015). Consequently, 463 infected trees will have a lower capacity to recover from drought, exacerbating the hydraulic 464 dysfunction and the NSC depletion. A common fungal pathogen that causes defoliation and 465 mortality in Q. ilex is Phytophthora cinnamomi (Sánchez et al. 2002). There is yet no 466 information available on the potential initiating or contributing role of this fungal pathogen in 467 drought-induced forest dieback of Q. ilex or during the recovery after these events. 468 Nonetheless, an experimental study conducted with Q. ilex seedlings demonstrated that plants 469 subjected to drought were more likely to die after being inoculated with this pathogen 470 (Corcobado et al. 2014). Interestingly, the most damaging treatment included a prior water- 471 logging treatment, followed first by the inoculation and then by drought treatment. Hence, the 472 sequence of the meteorological extremes matters, as it may interact with the activity of biotic 473 agents and amplify the damage of droughts. For this reason, it is required to conduct 474 experimental and field studies to understand the role of pathogens previously, during and 475 after drought-induced forest dieback of Q. ilex. 476 The role of forest structure in episodes of forest dieback 477 The spatial variation in growth and mortality rates of some Mediterranean tree species is 478 largely related to forest structure, more so than to climate (Vilà-Cabrera et al. 2011). Indeed, 479 the demographic response by forests to climate disturbances (e.g. extreme droughts) is 480 determined by structural characteristics, such as tree density and stand basal area (Lloret et al. 481 2012, Carnicer et al. 2014). The abandonment of forest management practices over the last 482 decades has resulted higher stand densities, which, in turn, increased competition for 483 resources (Vayreda et al. 2012). The widely distributed Quercus ilex forests are a good example 484 of this process. Coll et al. (2013) showed competition constraints on growth illustrated by 485 lower growth rates of Q. ilex in stands with high basal area, which is supported by less 486 noticeable effects of drought on tree water use in low density stands compared to high density 487 stands (Moreno and Cubera 2008). In the Iberian Peninsula, many forests are recovering from 488 overexploitation during the 1950s and 1960s, so they are relatively young forests that are 489 increasing in biomass (Vayreda et al. 2012). Along with the accumulation of biomass, there is a 490 progressive but slow self-thinning (López et al. 2009). Extreme droughts may partly accelerate 491 self-thinning by killing stems of multi-stemmed trees and by suppressing the smaller 492 individuals (Galiano et al. 2012; Barbeta et al. 2013). Consequently, the stands with high stem 493 density and basal area (and thus with a potential larger degree competition for water) are a 494 priori more prone to suffering large impacts during extreme drought episodes. These impacts 495 may involve extensive biomass loss and the death of individuals and thus, the surviving trees 496 will have less stems per individual, and the stand basal area will be concentrated in a reduced 497 number of individuals. It is likely therefore that the forest structure resulting from a drought 498 impact will present more resistance to drought episodes in the long-term. The convergence 499 towards structures adapted to the new, lower availability of water and more recurrent drought 500 episodes can be expected. Research addressed to predict which the most suitable forest 501 structure is for each sub-type of climate and topography should be a primary goal in order to 502 adapt management practices to the ongoing climate change. 503 Sequence of responses of Q. ilex to long-term field experimental drought 504 Two long-term drought experiments have been established in natural forests of Q. ilex; the 505 first in 1998 in Prades (north-eastern Iberian Peninsula) and the other in 2003 in the 506 Puébachon State Forest (southern France). Both experimental designs excluded throughfall 507 over ca. 30% of the plot surface (Limousin, Misson, et al. 2010), and runoff at the Prades site 508 was intercepted by ditches along the top edges of the plots (Ogaya and Peñuelas 2003). These 509 experiments have produced a large body of information on the responses of Q. ilex forests to 510 imposed moderate but continuous drought, from leaf-level physiology to ecosystem-level 511 demography and fluxes of energy and matter. In fact, the drought imposed by this throughfall 512 exclusion matches the predictions of increasingly drier conditions and the decrease in soil 513 moisture for the western Mediterranean Basin for the next decades (IPCC, 2013). Severe 514 drought treatments have often been applied to seedlings and/or saplings, providing a useful 515 mechanistic understanding of the sequence of responses to drought. The still relatively short 516 duration of these experiments, however, compromises the implementation of their results in 517 long-term coupled climate-vegetation models. Seedlings and adults from the same species 518 may also have different water-use strategies (Mediavilla and Escudero 2004), so the results of 519 experiments on seedlings may not be comparable with those on adults. Furthermore, the 520 strength of the effect on ecosystems of the same level of stress induced by an pulse type 521 experimental treatment (e.g. warming, drought, or nutrient addition) decreases over time 522 (Leuzinger et al. 2011). In contrast, the results of the long-term drought experiments in Prades 523 and Puébachon would identify more stable effects and responses that can be expected in 524 natural environments. In this section, we review the findings from these two experimental 525 systems (Table 3; Figure 5). 526 The removal of ca. 30% of the throughfall reduced soil moisture by an average of 13% 527 in the drought-treatment plots in Prades (Liu et al. 2015). An analogous reduction of ca. 28% 528 throughfall significantly increased water stress in the drought plots of Puébachon, as it induced 529 a consistent decrease in predawn leaf water potentials of droughted plants (Limousin et al. 530 2009, 2012, Limousin, Misson, et al. 2010). Predawn water potential is often considered a 531 surrogate of soil water potential (Ritchie and Hinckley 1975, Barbeta et al. 2012). Midday leaf 532 water potentials, though, did not generally differ between treatment and control (Limousin et 533 al. 2009, Ogaya, Barbeta, et al. 2014, Barbeta et al. 2015). As a short-term response to an 534 increase in xylem tension produced by lower soil-water potentials, plants down-regulated their 535 canopy conductance and daily transpiration rates (Limousin et al. 2009). Water stress caused 536 by lower amounts of soil moisture thus clearly entailed a decrease in the soil-plant- 537 atmosphere water flow (transpiration), shortly after the treatment had commenced (Limousin 538 et al. 2009), showing a rapid acclimation of plant funtioning to the new drier conditions. The 539 consistent decrease in total transpiration was not accompanied by significant modifications of 540 foliar gas exchange, neither the short term (after 1-4 years of drought treatment (Limousin, 541 Misson, et al. 2010, Ogaya, Llusià, et al. 2014), nor the long term (after 14 years (Ogaya, Llusià, 542 et al. 2014, Sperlich et al. 2015), indicating physiological homeostasis (Martin-Stpaul et al. 543 2013). It should be noted, however, that measurements of leaf-level gas exchange are 544 instantaneous. The overall decrease in transpiration may not be detectable by gas-exchange 545 analysers that measure at a scale of seconds. Daily (or a longer timescale) integrals of gas- 546 exchange measurements might be able to identify differences between treatments, as do 547 measurements of sap flow (Limousin et al. 2009). The concept of leaf-level physiological 548 homeostasis has been challenged in a recent study by the measurement of mesophyllic 549 conductance and daytime respiration (Sperlich et al. 2015). These authors have provided 550 evidence that mesophyllic conductance, in addition to stomatal conductance, can acclimate to 551 drought conditions by facilitating the diffusion of CO2 to chloroplasts and that, in the long 552 term, the droughted plants can increase the plasticity of their photosynthetic traits to take 553 advantage of environmentally favourable periods. 554 The decrease in transpiration reported by Limousin et al. (2009) may seem 555 inconsistent with the results of other drought treatments (Martin-StPaul et al. 2013, Ogaya, 556 Llusià, et al. 2014), but the droughted plants at both the Prades and Puébachon experimental 557 sites had reduced leaf areas after 3 years (Ogaya and Peñuelas 2006) and 7 years (Limousin et 558 al. 2012) of treatment. This reduction was not accompanied by a reduction in sapwood area, 559 and thus the ratio of leaf to sapwood area was decreased at the shoot level (Limousin, 560 Longepierre, et al. 2010, Limousin et al. 2012), indicating that tree-level transpiration was 561 reduced by decreasing the transpiring (leaf) area, which is also in agreement with the long- 562 term increase in leaf litterfall in the drought treatment observed in Prades (Liu et al 2015). 563 Indeed, leaf-specific conductivity has increased in the long term in droughted plants as a 564 consequence of a decrease in the ratio of leaf to sapwood area (Martin-StPaul et al. 2013). In 565 parallel with the changes in leaf to shoot allometry, individual leaf area and hydraulic 566 conductivity and stem anatomy and hydraulic properties have changed little. Root and branch 567 xylem vulnerability and stem specific conductivity did not differ significantly in the trees at 568 Puébachon after 6 years of the drought treatment (Limousin, Longepierre, et al. 2010). Vessel 569 diameter, frequency, and distribution also did not vary in the same study. Even though wood 570 anatomy is not necessarily associated with xylem vulnerability, the larger picture indicates that 571 the effects of mid- to long-term experimental drought are restricted to foliar anatomy, foliar 572 hydraulic properties and shoot-to-leaves allometry. 573 The acclimation of tree-level transpiration (Limousin et al. 2009), leaf area (Ogaya & 574 Peñuelas 2006), and leaf-to-shoot allometry (Limousin, Longepierre, et al. 2010) to drier 575 conditions entails a reduction in total photosynthetic area. Despite the lack of significant 576 differences in rates of carbon assimilation per unit leaf area, the amount of carbon assimilated 577 at the individual or stem level should be reduced. The amount of carbon allocated to growth 578 decreased in Prades after 2 years (Ogaya et al. 2003) and 5 years (Ogaya and Peñuelas 2007a) 579 of treatment, indicated by significantly lower rates of stem growth in the drought treatments. 580 Likewise, the allocation of carbon to reproduction was also reduced, indicated by a decrease in 581 fruit litterfall (Ogaya and Peñuelas 2007b). In contrast, stem growth was never affected in 582 Puébachon after 9 years of drought treatment. Indeed, a recent analysis of a longer data series 583 (11 years) in Prades found that the differences in stem growth tended to dampen over time 584 (Barbeta et al. 2013), which has been linked to a potential increase of resources as a 585 consequence of higher mortality rates in the drought plots (Barbeta et al. 2013, Lloret et al. 586 2012) and to epigenetic acclimation (Rico et al. 2013). That is, the relatively larger decrease in 587 stem density produced by a higher mortality in the drought plots has led to lower competition 588 for water. Furthermore, the physiological and morphological mechanisms of acclimation 589 described above may have provided to droughted plants a greater resistance to naturally 590 occurring extreme droughts, such as those in 2005-2006 and 2011 in Prades (Ogaya, Barbeta, 591 et al. 2014). Additionally, the self-thinning related to extreme drought and the drought 592 treatment kills those stems that had previously lower growth rates (Barbeta et al. 2013). 593 Consequently, stems with higher growth rates would have been selected by self-thinning in the 594 drought plots, influencing the difference between treatments. 595 The responses to drought after a decade or more after the onset of the experiments 596 differed slightly from those early in the experiments. Q. ilex has shown it to able to display 597 some mechanisms of acclimation to drought, which would be useful for its persistence under a 598 future drier climate. Overall forest productivity tended to decrease (Liu et al. 2015) with the 599 forests reducing their biomass in accordance with the artificially imposed ca. 30% reduction in 600 throughfall. Less dense forests with sparser canopies would thrive under the projected 15% 601 decrease in soil moisture in the Mediterranean Basin (IPCC 2013), but as we described above, 602 the effects of extreme droughts are more severe than those of drought experiments that 603 substantially reduce the water availability permanently. Because both an increase in the 604 frequency of extreme drought events and generally lower water availability are likely to occur 605 in the future, continuing investigating the interaction between extreme drought and the 606 acclimation mechanisms of Q. Ilex identified by the long-term experiments is necessary. One 607 could expect that a forest adapted to lower water availability would be more resistant to an 608 extreme drought. Trees suffering the effects of a long-term drought experiment, however, 609 stored similar NSC than the control trees (Rosas et al. 2013) so their resilience would be 610 equivalent after droughts that cause total canopy loss. In contrast, groundwater contributed 611 less to transpiration in the trees in the control plot during the drought of 2011 (after 12 years 612 of drought treatment), which indicated a relatively stronger depletion of groundwater reserves 613 (Barbeta et al. 2015). These reserves respond to the water balance over longer periods, so the 614 chronic water stress imposed by the experiment would prevent them from being ready and 615 available to minimise the negative effects of drought. 616 Concluding remarks 617 The objective of this review was to shed light on the key factors that determine potential 618 changes in the distribution, structure, and function of Q. ilex forests in response to increasing 619 droughts, either in the form of extreme events or as a sustained decrease in water availability. 620 This is a critical issue not only for Q. ilex forests but also for other ecosystems dominated by 621 resprouting angiosperms in all the Mediterranean regions of the world that share many of the 622 life-history traits and physiological mechanisms presented in Table 1. Analysing the responses 623 to droughts along different timescales, some of the key determinants of drought responses 624 have been described. Gaps in our knowledge that warrant further research, however, still 625 exist. From the dichotomy shown in Table 1 between developmental time and duration of 626 drought-adaptive traits or mechanisms, we can conclude that not all of these traits or 627 mechanisms could be similarly relevant in the future. In particular, the capacity to modify the 628 phenology of growth and carbon assimilation (Lempereur et al. 2015) is a promising 629 compensatory strategy for Q. ilex. In this sense, the limits of plasticity in primary and 630 secondary growth in relation to future drier and warmer conditions may be determinants for 631 the persistence of some populations in their current structure and function. We need to 632 determine if an increase in winter photosynthetic activity and an earlier spring budburst could 633 offset the deleterious effects of a longer summer cessation of growth (Carnicer et al. 2013). 634 This seasonal compensation is not only relevant for species sharing similar phenological 635 patterns and life-history traits with Quercus ilex such as Mediterranean woody angiosperms, 636 but also for winter- and drought-deciduous species in temperate and subtropical 637 environments. 638 The results of field drought experiments have indicated that trees are able to adjust 639 their tree- and stand-level leaf area to a lower availability of water. This adjustment increases 640 their resistance to drought, but we still do not know if they are also more resilient. The 641 depletion of NSC reserves in resprouting organs, such as lignotubers, indicates that the 642 capacity to resprout could be weakened. This weakened resprouting capacity has been 643 observed in the field following episodes of extreme drought. However, the permanent 644 reduction of water availability imposed by field drought experiments does not seem to reduce 645 lignotuber reserves and thus the resprouting capacity (Rosas et al. 2013). Still, this does not 646 directly imply that the resistance to extreme drought events associated with the current 647 climatic change remains unchanged after this permanent reductions of water availability. 648 Indeed, a drought in 2011 had a more negative impact on the drought plots (Barbeta et al. 649 2015), and a recent global-scale study evaluating the legacy of extreme droughts on trees has 650 estimated that growth could be reduced for 4 years after a single event (Anderegg et al. 2015). 651 In addition, the broad-scale analysis of the drought characteristics that triggered Q. ilex forest 652 dieback presented here shows that the duration of drought is the key parameter in 653 determining the severity of drought effects on Quercus ilex forests. The multi-year dynamics of 654 non-structural carbohydrates and/or the accumulated hydraulic dysfunction may be behind 655 the fact that persistent droughts cause more crown defoliation and mortality on Q. ilex forests 656 than droughts of short duration regardless of their intensity. 657 Individuals and plant communities have an extraordinary capacity to accommodate to 658 environmental changes. This is especially true for Quercus ilex and the forests that it 659 dominates. Such long-lived woody species have thrived in the same geographical area at least 660 since the Middle Würm (Reille and Pons 1992), facing contrasting climatic conditions. A 661 characteristic that may have allowed suach plants to survive these climatic periods is their 662 extensive root system, but when the access to below-ground water pools is impaired by long- 663 lasting droughts forest dieback may occur (Barbeta et al. 2015). To sum up, it is still not clear 664 whether the pace of the current climatic change can be matched by acclimation processes and 665 mechanisms showed here. More particularly, shorter periods of return of extreme droughts 666 may be decisive in determining the fate of Q. ilex forests and perhaps other Mediterranean 667 vegetation types. We recommend that future research should keep on addressing the 668 combined effect of consecutive extreme droughts and drier average conditions on the 669 structure and function of plant communities, but with a special emphasis on the resilience 670 after crown damage and on the access to the vital long-lasting deep water pools. 671 Acknowledgements 672 This research was financially supported by the ERC Synergy project SyG-2013-610028 673 IMBALANCE-P, by the Spanish Government project CGL2013-48074-P and by the Catalan 674 government project SGR2014-274. A.B. acknowledges an FPI predoctoral fellowship from the 675 Spanish Ministry of Economy and Competitiveness (BES-2011-043314). The authors thank Jofre 676 Carnicer, Víctor Resco de Dios, Laszlo Nagy and three anonymous reviewers for the insightful 677 comments to earlier versions of this manuscript. 678 679 References 680 Abrantes, J., Campelo, F., García-González, I., and Nabais, C., 2013. Environmental control of 681 vessel traits in Quercus ilex under Mediterranean climate: Relating xylem anatomy to 682 function. Trees - Structure and Function, 27, 655–662. 683 Aguade, D., Poyatos, R., Gomez, M., Oliva, J., and Martinez-Vilalta, J., 2015. The role of 684 defoliation and root rot pathogen infection in driving the mode of drought-related 685 physiological decline in Scots pine (Pinus sylvestris L.). Tree Physiology, 35 (Ipcc 2013), 686 229–242. 687 Aguadé, D., Poyatos, R., Rosas, T., and Martínez-vilalta, J., 2015. Comparative Drought 688 Responses of Quercus ilex L. and Pinus sylvestris L. in a Montane Forest Undergoing a 689 Vegetation Shift. Forests, 6, 2505–2529. 690 Albuixech, J., Camarero, J.J., and Montserrat-Marti, G., 2012. Dinámica estacional del 691 crecimiento secundario y anatomía del xilema en dos Quercus mediterráneos que 692 coexisten. Forest Systems, 21 (1), 9–22. 693 Alla, A.Q., Camarero, J.J., Maestro-Martínez, M., and Montserrat-Martí, G., 2012. Acorn 694 production is linked to secondary growth but not to declining carbohydrate 695 concentrations in current-year shoots of two oak species. Trees - Structure and Function, 696 26, 841–850. 697 Allen, C.D., Breshears, D.D., and McDowell, N.G., 2015. On underestimation of global 698 vulnerability to tree mortality and forest die-off from hotter drought in the 699 Anthropocene. Ecosphere, 6 (April), 1–55. 700 Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., 701 Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P., Fensham, R., 702 Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G., Running, S.W., Semerci, A., and 703 Cobb, N., 2010. A global overview of drought and heat-induced tree mortality reveals 704 emerging climate change risks for forests. Forest Ecology and Management, 259 (4), 660– 705 684. 706 Anderegg, W., Schwalm, C., Biondi, F., Camarero, J.J., Koch, G., Litvak, M., Ogle, K., Shaw, J.D., 707 Shevliakova, E., Williams, A.P., Wolf, A., Ziaco, E., and Pacala, S., 2015. Pervasive drought 708 legacies in forest ecosystems and their implications for carbon cycle models W. Science, 709 349 (6347). 710 Anderegg, W.R.L., Berry, J.A, and Field, C.B., 2012. Linking definitions, mechanisms, and 711 modeling of drought-induced tree death. Trends in plant science, 17 (12), 693–700. 712 Anderegg, W.R.L., Plavcová, L., Anderegg, L.D.L., Hacke, U.G., Berry, J.A., and Field, C.B., 2013. 713 Drought’s legacy: multiyear hydraulic deterioration underlies widespread aspen forest 714 die-off and portends increased future risk. Global Change Biology, 19, 1188–1196. 715 Babita, M., Maheswari, M., Rao, L.M., Shanker, A.K., and Rao, D.G., 2010. Osmotic adjustment, 716 drought tolerance and yield in castor (Ricinus communis L.) hybrids. Environmental and 717 Experimental Botany, 69 (3), 243–249. 718 Barbeta, A., Mejía-Chang, M., Ogaya, R., Voltas, J., Dawson, T.E., and Peñuelas, J., 2015. The 719 combined effects of a long-term experimental drought and an extreme drought on the 720 use of plant-water sources in a Mediterranean forest. Global Change Biology, 21, 1213– 721 1225. 722 Barbeta, A., Ogaya, R., and Peñuelas, J., 2012. Comparative study of diurnal and nocturnal sap 723 flow of Quercus ilex and Phillyrea latifolia in a Mediterranean holm oak forest in Prades ( 724 Catalonia , NE Spain ). Trees, 26, 1651–1659. 725 Barbeta, A., Ogaya, R., and Peñuelas, J., 2013. Dampening effects of long-term experimental 726 drought on growth and mortality rates of a Holm oak forest. Global change biology, 19 727 (10), 3133–44. 728 Bartlett, M.K., Scoffoni, C., and Sack, L., 2012. The determinants of leaf turgor loss point and 729 prediction of drought tolerance of species and biomes: A global meta-analysis. Ecology 730 Letters, 15, 393–405. 731 Beikircher, B. and Mayr, S., 2009. Intraspecific differences in drought tolerance and acclimation 732 in hydraulics of Ligustrum vulgare and Viburnum lantana. Tree Physiology, 29, 765–775. 733 Bucci, S.J., Scholz, F.G., Campanello, P.I., Montti, L., Jimenez-Castillo, M., Rockwell, F. a, Manna, 734 L. La, Guerra, P., Bernal, P.L., Troncoso, O., Enricci, J., Holbrook, M.N., and Goldstein, G., 735 2012. Hydraulic differences along the water transport system of South American 736 Nothofagus species: do leaves protect the stem functionality? Tree physiology, 32 (7), 737 880–93. 738 Camarero, J., Sangüesa-Barreda, G., and Vergarechea, M., 2015. Prior height, growth, and 739 wood anatomy differently predispose to drought-induced dieback in two Mediterranean 740 oak species. Annals of Forest Science. 741 Camarero, J.J., Franquesa, M., and Sangüesa-Barreda, G., 2015. Timing of Drought Triggers 742 Distinct Growth Responses in Holm Oak: Implications to Predict Warming-Induced Forest 743 Defoliation and Growth Decline. Forests, 6, 1576–1597. 744 Camarero, J.J., Olano, J.M., and Parras, A., 2010. Plastic bimodal xylogenesis in conifers from 745 continental Mediterranean climates. The New phytologist, 185 (2), 471–80. 746 Campelo, F., Nabais, C., Gutiérrez, E., Freitas, H., and García-González, I., 2010. Vessel features 747 of Quercus ilex L. growing under Mediterranean climate have a better climatic signal than 748 tree-ring width. Trees - Structure and Function, 24, 463–470. 749 Canadell, J., Djema, A., López, B., Lloret, F., Sabate, S., Siscart, D., and Gracia, C.A., 1999. 750 Structure and dynamics of the root system. In: F. Rodà, J. Retana, C.A. Gracia, and J. 751 Bellot, eds. Ecology of Mediterranean Evergreen Oak Forests. Springer Berlin / 752 Heidelberg, 47–59. 753 Canadell, J., Jackson, R.B., Ehleringer, J.B., Mooney, H.A., Sala, O.E., and Schulze, E.D., 1996. 754 Maximum rooting depth of vegetation types at the global scale. Oecologia, 108 (4), 583– 755 595. 756 Canadell, J. and López-Soria, L., 1998. Lignotuber reserves support regrowth following clipping 757 of two Mediterranean shrubs. Functional Ecology, 12, 31–38. 758 Carnicer, J., Barbeta, A., Sperlich, D., Coll, M., and Peñuelas, J., 2013. Contrasting trait 759 syndromes in angiosperms and conifers are associated with different responses of tree 760 growth to temperature on a large scale. Frontiers in plant science, 4 (October), 409. 761 Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., and Peñuelas, J., 2011. Widespread 762 crown condition decline, food web disruption, and amplified tree mortality with 763 increased climate change-type drought. Proceedings of the National Academy of Sciences 764 of the United States of America, 108 (4), 1474–1478. 765 Carnicer, J., Coll, M., Pons, X., Ninyerola, M., Vayreda, J., and Peñuelas, J., 2014. Large-scale 766 recruitment limitation in Mediterranean pines: The role of Quercus ilex and forest 767 successional advance as key regional drivers. Global Ecology and Biogeography, 23, 371– 768 384. 769 Cavender-Bares, J., Cortes, P., Rambal, S., Joffre, R., Miles, B., and Rocheteau, A., 2005. 770 Summer and winter sensitivity of leaves and xylem to minimum freezing temperatures: a 771 comparison of co-occurring Mediterranean oaks that differ in leaf lifespan. New 772 Phytologist, 168 (3), 597–612. 773 Choat, B., Jansen, S., Brodribb, T.J., Cochard, H., Delzon, S., Bhaskar, R., Bucci, S.J., Feild, T.S., 774 Gleason, S.M., Hacke, U.G., Jacobsen, A.L., Lens, F., Maherali, H., Martínez-Vilalta, J., 775 Mayr, S., Mencuccini, M., Mitchell, P.J., Nardini, A., Pittermann, J., Pratt, R.B., Sperry, J.S., 776 Westoby, M., Wright, I.J., and Zanne, A.E., 2012. Global convergence in the vulnerability 777 of forests to drought. Nature, 491 (7426), 752–5. 778 Coll, M., Peñuelas, J., Ninyerola, M., Pons, X., and Carnicer, J., 2013. Multivariate effect 779 gradients driving forest demographic responses in the Iberian Peninsula. Forest Ecology 780 and Management, 303, 195–209. 781 Comas, C., Castillo, J., Voltas, J., and Ferrio, J.P., 2015. Point processes statistics of stable 782 isotopes : analysing water uptake patterns in a mixed stand of Aleppo pine and Holm oak. 783 Forest Systems, 24 (1), 1–13. 784 Corcobado, T., Cubera, E., Juárez, E., Moreno, G., and Solla, A., 2014. Drought events 785 determine performance of Quercus ilex seedlings and increase their susceptibility to 786 Phytophthora cinnamomi. Agricultural and Forest Meteorology, 192-193, 1–8. 787 Corcuera, L., Camarero, J.J., and Gil-Pelegrín, E., 2004. Effects of a severe drought on Quercus 788 ilex radial growth and xylem anatomy. Trees - Structure and Function, 18, 83–92. 789 David, T.S., Ferreira, M.I., Cohen, S., Pereira, J.S., and David, J.S., 2004. Constraints on 790 transpiration from an evergreen oak tree in southern Portugal. Agricultural and Forest 791 Meteorology, 122, 193–205. 792 Delzon, S., Urli, M., Samalens, J.C., Lamy, J.B., Lischke, H., Sin, F., Zimmermann, N.E., and Porté, 793 A.J., 2013. Field evidence of colonisation by holm oak, at the Northern Margin of its 794 distribution range, during the anthropocene period. PLoS ONE, 8 (11), 1–8. 795 Deslauriers, A., Rossi, S., Anfodillo, T., and Saracino, A., 2008. Cambial phenology, wood 796 formation and temperature thresholds in two contrasting years at high altitude in 797 southern Italy. Tree physiology, 28, 863–871. 798 Desprez-Loustau, M.-L., Marçais, B., Nageleisen, L.-M., Piou, D., and Vannini, A., 2006. 799 Interactive effects of drought and pathogens in forest trees. Annals of Forest Science, 63, 800 597–612. 801 Espelta, J., Sabaté, S., and Retana, J., 1999. Resprouting dynamics. In: Ecology of 802 Mediterranean Evergreen Oak Forests. Berlin: Springer, 61–73. 803 Ferrio, J.P., Florit, A., Vega, A., Serrano, L., and Voltas, J., 2003. Delta(13)C and tree-ring width 804 reflect different drought responses in Quercus ilex and Pinus halepensis. Oecologia, 137 805 (4), 512–518. 806 Galiano, L., Martínez-Vilalta, J., Sabaté, S., and Lloret, F., 2012. Determinants of drought effects 807 on crown condition and their relationship with depletion of carbon reserves in a 808 Mediterranean holm oak forest. Tree Physiology, 32 (4), 478–489. 809 Gea-Izquierdo, G., Martín-Benito, D., Cherubini, P., and Cañellas, I., 2009. Climate-growth 810 variability in Quercus ilex L. west Iberian open woodlands of different stand density. 811 Annals of Forest Science, 66 (8), 802. 812 Gratani, L., Meneghini, M., Pesoli, P., and Crescente, M.F., 2003. Structural and functional 813 plasticity of Quercus ilex seedlings of different provenances in Italy. Trees - Structure and 814 Function, 17, 515–521. 815 Gutiérrez, E., Campelo, F., Camarero, J., Ribas, M., Muntán, E., Nabais, C., and Freitas, H., 2011. 816 Climate controls act at different scales on the seasonal pattern of Quercus ilex L. stem 817 radial increments in NE Spain. Trees Structure and Function, 25 (4), 637–646. 818 Hacke, U.G., Stiller, V., Sperry, J.S., Pittermann, J., and McCulloh, K.A., 2001. Cavitation Fatigue. 819 Embolism and Refilling Cycles Can Weaken the Cavitation Resistance of Xylem. Plant 820 Physiology, 125, 779–786. 821 Hernández, E.I., Pausas, J.G., and Vilagrosa, A., 2011. Leaf physiological traits in relation to 822 resprouter ability in the Mediterranean Basin. Plant Ecology, 212, 1959–1966. 823 IPCC, 2013. Climate Change 2013: The Physical Science Basis. Contribution of Working Group I 824 to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. 825 Cambridge, UK. New York, NY, USA: Cambridge University Press. 826 Jentsch, A., Kreyling, J., and Beierkuhnlein, C., 2007. A new generation of events , not trends 827 experiments : Frontiers in Ecology and the Environment, 5, 365–374. 828 Klimešová, J. and Klimeš, L., 2007. Bud banks and their role in vegetative regeneration - A 829 literature review and proposal for simple classification and assessment. Perspectives in 830 Plant Ecology, Evolution and Systematics, 8 (1977), 115–129. 831 Körner, C., 2003. Carbon limitation in trees. Journal of Ecology, 91 (1), 4–17. 832 De la Cruz, A.C., Gil, P.M., Fernández-Cancio, Á., Minaya, M., Navarro-Cerrillo, R.M., Sánchez- 833 Salguero, R., and Grau, J.M., 2014. Defoliation triggered by climate induced effects in 834 Spanish ICP Forests monitoring plots. Forest Ecology and Management, 331 835 (NOVEMBER), 245–255. 836 Lempereur, M., Martin-stpaul, N.K., Damesin, C., Joffre, R., Ourcival, J., Rocheteau, A., and 837 Rambal, S., 2015. Growth duration is a better predictor of stem increment than carbon 838 supply in a Mediterranean oak forest : implications for assessing forest productivity 839 under climate change. New Phytologist, 207, 579–590. 840 Leuzinger, S., Luo, Y., Beier, C., Dieleman, W., Vicca, S., and Körner, C., 2011. Do global change 841 experiments overestimate impacts on terrestrial ecosystems? Trends in ecology & 842 evolution, 26 (5), 236–41. 843 Limousin, J.-M., Longepierre, D., Huc, R., and Rambal, S., 2010. Change in hydraulic traits of 844 Mediterranean Quercus ilex subjected to long-term throughfall exclusion. Tree 845 physiology, 30, 1026–1036. 846 Limousin, J.-M., Misson, L., Lavoir, A., and Martin, N.K., 2010. Do photosynthetic limitations of 847 evergreen Quercus ilex leaves change with long-term increased drought severity ? Plant 848 Cell and Environment, 33, 863–875. 849 Limousin, J.-M., Rambal, S., Ourcival, J.-M., Rocheteau, A., Joffre, R., and Rodríguez-Cortina, R., 850 2009. Long-term transpiration change with rainfall decline in a Mediterranean Quercus 851 ilex forest. Global Change Biology, 15, 2163–2175. 852 Limousin, J.-M., Rambal, S., Ourcival, J.-M., Rodríguez-Calcerrada, J., Pérez-Ramos, I.M., 853 Rodríguez-Cortina, R., Misson, L., and Joffre, R., 2012. Morphological and phenological 854 shoot plasticity in a Mediterranean evergreen oak facing long-term increased drought. 855 Oecologia, 169 (2), 565–77. 856 Liu, D., Ogaya, R., Barbeta, A., Yang, X., and Peñuelas, J., 2015. Contrasting impacts of 857 continuous moderate drought and episodic severe droughts on the aboveground- 858 biomass increment and litterfall of three coexisting Mediterranean woody species. Global 859 Change Biology, n/a–n/a. 860 Lloret, F., Escudero, A., Iriondo, J.M., Martínez-Vilalta, J., and Valladares, F., 2012. Extreme 861 climatic events and vegetation: the role of stabilizing processes. Global Change Biology, 862 18 (3), 797–805. 863 Lloret, F., Siscart, D., and Dalmases, C., 2004. Canopy recovery after drought dieback in holm- 864 oak Mediterranean forests of Catalonia (NE Spain). Global Change Biology, 10 (12), 2092– 865 2099. 866 López, B.C., Gracia, C. a., Sabaté, S., and Keenan, T., 2009. Assessing the resilience of 867 Mediterranean holm oaks to disturbances using selective thinning. Acta Oecologica, 35, 868 849–854. 869 Martínez-Vilalta, J., Mangirón, M., Ogaya, R., Sauret, M., Serrano, L., Peñuelas, J., and Piñol, J., 870 2003. Sap flow of three co-occurring Mediterranean woody species under varying 871 atmospheric and soil water conditions. Tree physiology, 23 (11), 747–58. 872 Martínez-Vilalta, J., Poyatos, R., Aguadé, D., Retana, J., and Mencuccini, M., 2014. A new look 873 at water transport regulation in plants. New Phytologist, 204, 105–115. 874 Martínez-Vilalta, J., Prat, E., Oliveras, I., and Piñol, J., 2002. Xylem hydraulic properties of roots 875 and stems of nine Mediterranean woody species. Oecologia, 133, 19–29. 876 Martin-Stpaul, N.K., Limousin, J.-M., Vogt-Schilb, H., Rodríguez-Calcerrada, J., Rambal, S., 877 Longepierre, D., and Misson, L., 2013. The temporal response to drought in a 878 Mediterranean evergreen tree: comparing a regional precipitation gradient and a 879 throughfall exclusion experiment. Global Change Biology, 19 (8), 2413–26. 880 Matesanz, S. and Valladares, F., 2013. Ecological and evolutionary responses of Mediterranean 881 plants to global change. Environmental and Experimental Botany. 882 Mcadam, S.A., M. and Brodribb, T.J., 2012. Stomatal innovation and the rise of seed plants. 883 Ecology Letters, 15, 1–8. 884 Mediavilla, S. and Escudero, A., 2004. Stomatal responses to drought of mature trees and 885 seedlings of two co-occurring Mediterranean oaks. Forest Ecology and Management, 187, 886 281–294. 887 Meinzer, F.C. and McCulloh, K. a, 2013. Xylem recovery from drought-induced embolism: 888 where is the hydraulic point of no return? Tree physiology, 33 (4), 331–4. 889 Mitrakos, K., 1980. A theory for Mediterranean plant life. Acta Oecologica, (1), 245–252. 890 Montserrat-Martí, G., Camarero, J.J., Palacio, S., Pérez-Rontomé, C., Milla, R., Albuixech, J., and 891 Maestro, M., 2009. Summer-drought constrains the phenology and growth of two 892 coexisting Mediterranean oaks with contrasting leaf habit: Implications for their 893 persistence and reproduction. Trees - Structure and Function, 23, 787–799. 894 Mooney, H. a and Dunn, E.L., 1970. Convergent Evolution of Mediterranean-Climate Evergreen 895 Sclerophyll Shrubs. Evolution, 24 (2), 292–303. 896 Moreno, G. and Cubera, E., 2008. Impact of stand density on water status and leaf gas 897 exchange in Quercus ilex. Forest Ecology and Management, 254 (1), 74–84. 898 Nardini, A., Lo Gullo, M.A., and Salleo, S., 2011. Refilling embolized xylem conduits: is it a 899 matter of phloem unloading? Plant science : an international journal of experimental 900 plant biology, 180 (4), 604–11. 901 Niinemets, Ü., 2015. Is there a species spectrum within the world-wide leaf economics 902 spectrum ? Major variations in leaf functional traits in the Mediterranean sclerophyll 903 Quercus ilex. New Phytologist, 205, 79–96. 904 Ogaya, R., Barbeta, A., Başnou, C., and Peñuelas, J., 2014. Satellite data as indicators of tree 905 biomass growth and forest dieback in a Mediterranean holm oak forest. Annals of Forest 906 Science, 135–144. 907 Ogaya, R., Llusià, J., Barbeta, A., Asensio, D., Liu, D., Alessio, G.A., and Peñuelas, J., 2014. Foliar 908 CO2 in a holm oak forest subjected to 15 years of climate change simulation. Plant 909 Science, 226, 101–107. 910 Ogaya, R. and Peñuelas, J., 2003. Comparative field study of Quercus ilex and Phillyrea latifolia: 911 photosynthetic response to experimental drought conditions. Environmental and 912 Experimental Botany, 50 (2), 137–148. 913 Ogaya, R. and Peñuelas, J., 2006. Contrasting foliar responses to drought in Quercus ilex and 914 Phillyrea latifolia. Biologia Plantarum, 50 (3), 373–382. 915 Ogaya, R. and Peñuelas, J., 2007a. Tree growth , mortality , and above-ground biomass 916 accumulation in a holm oak forest under a five-year experimental field drought. Plant 917 Ecology, 189, 291–299. 918 Ogaya, R. and Peñuelas, J., 2007b. Species-specific drought effects on flower and fruit 919 production in a Mediterranean holm oak forest. Forestry. 920 Ogaya, R., Peñuelas, J., Martínez-Vilalta, J., and Mangirón, M., 2003. Effect of drought on 921 diameter increment of Quercus ilex , Phillyrea latifolia , and Arbutus unedo in a holm oak 922 forest of NE Spain. Forest Ecology and Management, 180, 175–184. 923 Olano, J.M., Almería, I., Eugenio, M., and von Arx, G., 2013. Under pressure: How a 924 Mediterranean high-mountain forb coordinates growth and hydraulic xylem anatomy in 925 response to temperature and water constraints. Functional Ecology, 27, 1295–1303. 926 Oliva, J., Stenlid, J., and Martínez-Vilalta, J., 2014. The effect of fungal pathogens on the water 927 and carbon economy of trees : implications for drought-induced mortality included in 928 theoretical models for drought induced mortality. New Phytologist, 203 (C), 1028–1035. 929 Pasho, E., Camarero, J.J., de Luis, M., and Vicente-Serrano, S.M., 2011. Impacts of drought at 930 different time scales on forest growth across a wide climatic gradient in north-eastern 931 Spain. Agricultural and Forest Meteorology, 151 (12), 1800–1811. 932 Patakas, A., Nikolaou, N., Zioziou, E., Radoglou, K., and Noitsakis, B., 2002. The role of organic 933 solute and ion accumulation in osmotic adjustment in drought-stressed grapevines. Plant 934 Science, 163, 361–367. 935 Peguero-Pina, J.J., Sancho-Knapik, D., Barrón, E., Camarero, J.J., Vilagrosa, A., and Gil-Pelegrín, 936 E., 2014. Morphological and physiological divergences within Quercus ilex support the 937 existence of different ecotypes depending on climatic dryness. Annals of Botany, 114, 938 301–313. 939 Peñuelas, J., Filella, I., Lloret, F., Piñol, J., and Siscart, D., 2000. Effects of a severe drought on 940 water and nitrogen use by Quercus ilex and Phillyrea latifolia. Biologia Plantarum, 43 (1), 941 47–53. 942 Piñol, J., Lledó, M.J., and Escarré, A., 1991. Hydrological balance of two Mediterranean 943 forested catchments (Prades, northeast Spain). Hydrological Sciences Journal, 36 944 (September 2015), 95–107. 945 Pivovaroff, A.L., Sack, L., and Santiago, L.S., 2014. Coordination of stem and leaf hydraulic 946 conductance in southern California shrubs: A test of the hydraulic segmentation 947 hypothesis. New Phytologist, 203, 842–850. 948 Poyatos, R., Aguadé, D., Galiano, L., Mencuccini, M., and Martínez-Vilalta, J., 2013. Drought- 949 induced defoliation and long periods of near-zero gas exchange play a key role in 950 accentuating metabolic decline of Scots pine. New Phytologist, 200 (2), 388–401. 951 Reichstein, M., Tenhunen, J.D., Roupsard, O., Rambal, S., Miglietta, F., and Peressotti, A., 2002. 952 Severe drought effects on ecosystem CO2 and H2O fluxes at three Mediterranean 953 evergreen sites: revision of current hypotheses? Global Change Biology, 8 (March), 999– 954 1017. 955 Reille, M. and Pons, A., 1992. The ecological significance of slerophyllous oak forests in the 956 western part of the Mediterranean basin: a note on pollen analytical data. In: F. Romane 957 and J. Terradas, eds. Quercus ilex L. ecosystems: function, dynamics and management. 958 Dordrecht: Kluwer, 13–18. 959 Rico, L., Ogaya, R., Barbeta, A., and Peñuelas, J., 2013. Changes in DNA methylation fingerprint 960 of Quercus ilex trees in response to experimental field drought simulating projected 961 climate change. Plant Biology. 962 Ritchie, G.A. and Hinckley, T.M., 1975. The Pressure Chamber as an Instrument for Ecological 963 Research. Advances in Ecological Research, 9, 165–254. 964 Rivas-Ubach, A., Gargallo-Garriga, A., Sardans, J., Oravec, M., Mateu-Castell, L., Pérez-Trujillo, 965 M., Parella, T., Ogaya, R., Urban, O., and Peñuelas, J., 2014. Drought enhances folivory by 966 shifting foliar metabolomes in Quercus ilex trees. New Phytologist, 202, 874–885. 967 Rosas, T., Galiano, L., Ogaya, R., Peñuelas, J., and Martínez-Vilalta, J., 2013. Dynamics of non- 968 structural carbohydrates in three Mediterranean woody species following long-term 969 experimental drought. Frontiers in Plant Science, 4 (October), 1–16. 970 Sala, A., Woodruff, D.R., and Meinzer, F.C., 2012. Carbon dynamics in trees: feast or famine? 971 Tree Physiology, 32 (6), 1–12. 972 Sánchez, M.E., Caetano, P., Ferraz, J., and Trapero, A., 2002. Phytophtora disease of Quercus 973 ilex in south-western Spain. Forest Pathology, 32, 5–18. 974 Sánchez-Costa, E., Poyatos, R., and Sabaté, S., 2015. Contrasting growth and water use 975 strategies in four co-occurring Mediterranean tree species revealed by concurrent 976 measurements of sap flow and stem diameter variations. Agricultural and Forest 977 Meteorology, 207, 24–37. 978 Smith, M.D., 2011. An ecological perspective on extreme climatic events: A synthetic definition 979 and framework to guide future research. Journal of Ecology, 99, 656–663. 980 Sperlich, D., Barbeta, A., Ogaya, R., Sabate, S., and Penuelas, J., 2015. Balance between carbon 981 gain and loss under long-term drought: impacts on foliar respiration and photosynthesis 982 in Quercus ilex L. Journal of Experimental Botany. 983 Terradas, J., 1999. Holm Oak and Holm Oak Forests: An Introduction. In: F. Rodà, J. Retana, C. 984 Gracia, and J. Bellot, eds. Ecology of Mediterranean Evergreen Oak Forests. Berlin: 985 Springer Berlin / Heidelberg, 3–14. 986 Trifilò, P., Barbera, P.M., Raimondo, F., Nardini, A., Lo Gullo, M.A., and Meinzer, F., 2014. 987 Coping with drought-induced xylem cavitation: Coordination of embolism repair and ionic 988 effects in three Mediterranean evergreens. Tree Physiology, 34 (i), 109–122. 989 Vayreda, J., Martinez-Vilalta, J., Gracia, M., and Retana, J., 2012. Recent climate changes 990 interact with stand structure and management to determine changes in tree carbon 991 stocks in Spanish forests. Global Change Biology, 18, 1028–1041. 992 Vicente-Serrano, S.M., Beguería, S., and López-Moreno, J.I., 2010. A multiscalar drought index 993 sensitive to global warming: The standardized precipitation evapotranspiration index. 994 Journal of Climate, 23, 1696–1718. 995 Vicente-Serrano, S.M., Gouveia, C., Camarero, J.J., Beguería, S., Trigo, R., López-Moreno, J.I., 996 Azorín-Molina, C., Pasho, E., Lorenzo-Lacruz, J., Revuelto, J., Morán-Tejeda, E., and 997 Sanchez-Lorenzo, A., 2013. Response of vegetation to drought time-scales across global 998 land biomes. Proceedings of the National Academy of Sciences of the United States of 999 America, 110, 52–7. 1000 Vieira, J., Rossi, S., Campelo, F., Freitas, H., and Nabais, C., 2014. Xylogenesis of Pinus pinaster 1001 under a Mediterranean climate. Annals of Forest Science, 71, 71–80. 1002 Vilà-Cabrera, A., Martínez-Vilalta, J., Vayreda, J., and Retana, J., 2011. Structural and climatic 1003 determinants of demographic rates of Scots pine forests across the Iberian Peninsula. 1004 Ecological applications : a publication of the Ecological Society of America, 21 (4), 1162– 1005 72. 1006 Vilagrosa, A., Hernández, E.I., Luis, V.C., Cochard, H., and Pausas, J.G., 2014. Physiological 1007 differences explain the co-existence of different regeneration strategies in 1008 Mediterranean ecosystems. New Phytologist, 201, 1277–1288. 1009 Villar-Salvador, P., Castro-Díez, P., Pérez-Rontomé, C., and Montserrat-Martí, G., 1997. Stem 1010 xylem features in three Quercus (Fagaceae) species along a climatic gradient in NE Spain. 1011 Trees, 12, 90–96. 1012 Van der Werf, G.W., Sass-Klaassen, U.G.W., and Mohren, G.M.J., 2007. The impact of the 2003 1013 summer drought on the intra-annual growth pattern of beech (Fagus sylvatica L.) and oak 1014 (Quercus robur L.) on a dry site in the Netherlands. Dendrochronologia, 25, 103–112. 1015 Wolkovich, E.M., Cook, B.I., McLauchlan, K.K., and Davies, T.J., 2014. Temporal ecology in the 1016 Anthropocene. Ecology letters, 17, 1365–1379. 1017 World Meteorological Organization, 1992. International Meteorlogical Vocabulary. Geneva, 1018 No. 182. 1019 Worrall, J.J., Marchetti, S.B., Egeland, L., Mask, R.A., Eager, T., and Howell, B., 2010. Effects and 1020 etiology of sudden aspen decline in southwestern Colorado, USA. Forest Ecology and 1021 Management, 260 (5), 638–648. 1022 Zeppel, M.J.B., Harrison, S.P., Adams, H.D., Kelley, D.I., Li, G., Tissue, D.T., Dawson, T.E., 1023 Fensham, R., Belinda, E., Palmer, A., West, A.G., and Mcdowell, N.G., 2015. Drought and 1024 resprouting plants. New Phytologist, 206, 583–589. 1025 1026 1027 Table 1. Adaptive morphological, physiological and phenological mechanisms and/or traits of Quercus ilex for withstanding summer droughts. Table 2. Characteristics of reported drought-induced declines in Quercus ilex forests involving crown defoliation and/or stem mortality. CD, crown defoliation. Table 3. Summary of the effects identified by studies of long-term experimental drought in Quercus ilex forests. Figure captions Figure 1. Monthly average difference between the monthly precipitation and the monthly potential evapotranspiration (the climatic water balance) for two forests dominated by Quercus ilex. Data was obtained from meteorological stations located at Torners Valley in the Prades Mountains (930 m a.s.l; 1998-2014) and at Viladrau in the Montseny Mountains (950 m a.s.l. 1995-2014). Potential Evapotranspiration is calculated with the SPEI package in R, based on monthly temperature and incoming solar radiation (Thornthwaite 1948). Error bars are the standard errors of the means. Figure 2. Map of the current distributional range of holm oak (Quercus ilex) in the Mediterranean Basin. Figure 3. Correlation heatmap between the annual relative change in crown defoliation and the SPEI values for Iberian Quercus ilex forests (1987-2013, 233 plots). The SPEI duration on the x-axis represents the number of months used to calculate the water balance (e.g. an SPEI duration of 3 in June is calculated using the water balance of the previous April and May and the June itself). The y-axis is the month in which the calculation of SPEI departs from. The panel on the right top indicates the correspondence of the colours in the heatmap with the Person’s r coefficient of the correlation between SPEI and defoliation. Figure 4. Time courses of (a) accumulated crown defoliation and (b) annual change in crown defoliation relative to defoliation from the previous years (red lines). Blue bars are the June SPEIs calculated at a timescale of 23 months, which was the best correlated index. Error bars are the standard errors of the means (N=233). Figure 5. Conceptual scheme of the most relevant effects of the long-term drought experiments in Quercus ilex forests at the tree and ecosystemic levels. The span of the rectangles defines the duration of the effects. The rectangles that surpass 12 years indicate effects that lasted till the present day of the experiment. The position on the x-axis represents the approximate time of appearance. Rectangles with dashed contour lines indicate potential effects that require further study. Figure 1 Figure 2 Figure 3 Figure 4 Figure 5