1   Topsoil depth substantially influences the responses to drought of the foliar metabolomes of
 2   Mediterranean forests
 3
 4   Albert Rivas-Ubach1,2,3, Adrià Barbeta2,3, Jordi Sardans2,3, Alex Guenther4, Romà Ogaya2,3,
 5   Michal Oravec5, Otmar Urban5, Josep Peñuelas2,3
 6

 7   1. Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA,
 8   USA, 99354
 9   2. CREAF, Global Ecology Unit CRAF-CSIC-UAB, Cerdanyola del Vallès, 08913 Catalonia, Spain
10   3. CSIC, Global Ecology Unit CREAF-CSIC-UAB, Cerdanyola del Vallès, 08913 Catalonia, Spain
11   4. Department of Earth System Science, University of California, Irvine, CA, USA 92697
12   5. Global Change Research Centre, Academy of Sciences of the Czech Republic, Bĕlidla 4a, CZ-603 00
13   Brno, Czech Republic
14
15   Author e-mail addresses:
16   Albert Rivas-Ubach: albert.rivas.ubach@gmail.com
17   Adrià Barbeta: a.barbeta@creaf.uab.cat
18   Jordi Sardans: j.sardans@creaf.uab.cat
19   Alex Guenther: alex.guenther@uci.edu
20   Romà Ogaya: r.ogaya@creaf.uab.cat
21   Michal Oravec: oravec.m@czechglobe.cz
22   Otmar Urban: urban.o@czechglobe.cz
23   Josep Peñuelas: josep.penuelas@uab.cat

24
25
26   Corresponding Author:
27   Albert Rivas-Ubach
28   Environmental Molecular Sciences Laboratory,
29   Pacific Northwest National Laboratory,
30   Richland, WA, USA, 99354
31   Telf: +1 509 371 7319
32   e-mail: albert.rivas.ubach@gmail.com
33
34   Abbreviations
35   H-Forest: Forest with high canopies
36   L-Forest: Forest with low canopies
37   VS: Vegetation structure
38   DUSL: Depth of upper soil layers
39   D50: Diameter at 50 cm from soil
40   SEM: Structural Equation Model
41   LC-MS: Liquid chromatography coupled to mass spectrometry
42
43
44

                                                       1
45   Abstract
46   The upper soil provides support, water, and nutrients to terrestrial plants and is therefore
47   crucial for forest dynamics. We hypothesised that a tree's metabolic activity (and therefore its
48   metabolome; the total set of metabolites) would be affected by both the depth of upper soil
49   layers and water availability. We sampled leaves for stoichiometric and metabolomic analyses
50   once per season from differently sized Quercus ilex trees under natural and experimental
51   drought conditions representing the likely conditions in the coming decades). Although the
52   metabolomes varied according to tree size, smaller trees did not show higher concentrations
53   of biomarker metabolites related to drought stress. However, the effect of the drought
54   treatment on the metabolomes was greatest for small trees growing in shallow soils. Our
55   results suggest that tree size is more dependent on the depth of the upper soil, which
56   indirectly affects a tree's metabolome, rather than on the moisture content in the upper soil.
57   Metabolomic profiling of Q. ilex supports our finding that water availability in the upper soil is
58   not necessarily correlated with tree size. The higher impact of drought on trees growing in
59   shallower soils nevertheless indicates that any increase in the frequency, intensity, and
60   duration of drought - as has been projected for the Mediterranean Basin and other areas -
61   would affect small trees most. Metabolomics has proved to be a useful means for investigating
62   the links between plant metabolism and environmental conditions.

63
64   Keywords: Vegetation structure, soil depth, soil moisture, metabolomics, Quercus ilex

65

66
67
68
69
70
71
     This is the author’s version of a work that was accepted for publication in Perspectives
72   in plant ecology and evolution (Ed. Elsevier) . Changes resulting from the publishing
     process, such as peer review, editing, corrections, structural formatting, and other
73
     quality control mechanisms may not be reflected in this document. Changes may have
74   been made to this work since it was submitted for publication. A definitive version was
     subsequently published in Rivas, A. et al “Topsoil depth substantially influences the
75   responses to drought of the foliar metabolomes of Mediterranean forests ” in
76   Perspectives in plant ecology, evolution and systematics, vol. 21 (Aug. 2016), p. 41-54.

77   DOI 10.1016/j.ppees.2016.06.001


                                                      2
 78   Introduction
 79   Soil provides a physical support system and a reservoir of water for terrestrial primary
 80   producers (Montheith, 1981). A scarcity of soil resources, particularly water, is often
 81   associated with restricted development of plant-soil systems and reduced biomass (Huxman et
 82   al., 2004; Knapp and Smith, 2001; Orwig and Abrams, 1997). Soil biological activity and tree
 83   growth can be limited by several factors such as nutrients (Bowman et al., 1993; Sardans and
 84   Peñuelas, 2015; Sardans et al., 2012a, 2012b), light (Poorter, 1999), temperature (Epstein et
 85   al., 1997), or water (Huxman et al., 2004; Rosenzweig, 1968; Sala et al., 1988). Topographic
 86   factors such as slope variation and/or soil texture also play important roles in the retention
 87   and storage of soil water (Farahani et
 88   al., 1998; Fernandez-Illescas et al., 2001) and can influence soil enzymatic activity (Bastida et
 89   al., 2008) and erosion (Kinnell and Cummings, 1993). Soil depth is tightly linked with the
 90   physiology and respiration rates of tree roots (Pregitzer et al., 1998), the composition of
 91   microbial communities (Fierer et al., 2003), and even plant biodiversity (Fuhlendorf and
 92   Smeins, 1998). Most of the biological activity and nutrient recycling in soil occurs in the upper
 93   topsoil layers, so water availability in these layers is crucial to forests (Jobbagy and Jackson,
 94   2000; Wardle et al., 2004). Hydraulic lift, mainly the transport of water from deep to shallower
 95   soil layers through roots to maintain physiological activity (Canadell et al., 1996; Nepstad et al.,
 96   1994; Schulze et al., 1996), is common in several plant species (Caldwell and Richards, 1989;
 97   Caldwell et al., 1998; Peñuelas and Filella, 2003; Prieto et al., 2012; Wan et al., 2000). Plants
 98   also intensely influence soil; interdependence between changes in plant communities and in
 99   soil properties, such as fertility, has been observed in the Mediterranean Basin, especially in
100   dry areas (Ruiz-Sinoga et al., 2011). For example, increased plant cover had a direct effect on
101   soil porosity by increasing water infiltration and decreasing runoff (Garcia-Estringana et al.,
102   2010; Goberna et al., 2007; Johnson-Maynard et al., 2002). The enhancement of soil quality by
103   plant cover thus improves fertility (Gallardo et al., 2000) and enzymatic activity (Garcia et al.,
104   2002), which in turn have positive effects on plant metabolism and growth (Ruiz-Sinoga et al.,
105   2011; Sadaka and Ponge, 2003). The depth and texture of the topsoil, amongst other
106   topographic factors, and the link between vegetation and soil are thus fundamental to the
107   availability of water and nutrients for plants and may further play crucial roles in determining
108   plant cover, habitat fragmentation, landscape patchiness, and changes in biodiversity (Fahrig,
109   2003; Sardans and Peñuelas, 2014; Allen and Breshears, 1998).
110       Mediterranean and other arid and semi-arid ecosystems have marked seasonality, with
111   hot and dry summers (Aschmann 1973; Schwinning et al., 2004). Soil depth can play a critical
112   role in these ecosystems, because summer drought is an important factor limiting the growth

                                                       3
113   of plants (Ogaya et al., 2003) and determining tree mortality (Barbeta et al., 2013). High
114   temperatures and the absence of precipitation during summer deplete moisture in the upper
115   soil layers, so plant activity can only be sustained where enough moisture is available in deep
116   soil layers (Barbeta et al., 2015). Mortality and crown damage caused by extreme droughts are
117   usually minimised where soil is deep (Lloret et al., 2012). The most recent climatic models
118   project more frequent and severe droughts in Mediterranean ecosystems (IPCC, 2007). The
119   effects of the increased drought conditions have already been observed during the last two
120   decades affecting forest communities by increasing crown defoliation and tree mortality (Allen
121   et al., 2010; Bigler et al., 2006; Carnicer et al., 2011; Galiano et al., 2011, 2010; McDowell et al.,
122   2008; Poyatos et al., 2013; Rebetez and Dobbertin, 2004). The effects of drought on plant
123   fitness and performance may also trigger important cascade effects through various trophic
124   levels (Harrison, 2001; Kuske et al., 2003), thus producing important changes to entire
125   ecosystems. A future exacerbation of drought in Mediterranean and other arid and semi-arid
126   ecosystems may thus potentially affect, in a more intense or different way, the metabolism of
127   trees growing in shallow soils with low capacities to store water, thereby leading to landscape
128   patchiness and desertification (Sardans and Peñuelas, 2014). Forest decline and/or large
129   changes in the structure of the vegetation and habitat are thus more likely in areas with
130   shallower soil that are more susceptible to extreme drought (Galiano et al., 2012; Lloret et al.,
131   2004).
132       Plants can adjust their metabolisms to maintain homeostasis under marked seasonality
133   (Bertram et al., 2010; Falasca et al., 2013; Rivas-Ubach et al., 2014, 2012). Extreme droughts
134   can cause drastic vegetation shifts, especially in Mediterranean, semi-arid, and arid
135   ecosystems (Allen and Breshears, 1998; Hanson and Weltzin, 2000; Mueller et al., 2005), so
136   the study of plant metabolomes can contribute to our understanding of how plants can
137   metabolically cope with intense drought stress. Plants under drought conditions can adjust
138   their chemistry to maintain physiological functions by, for example, increasing foliar
139   concentrations of K (Cakmak and Engels, 1999; Sardans and Peñuelas, 2015), proline (Tymms
140   and Gaff, 1979; Yamada et al., 2005), antioxidants (Rivas-Ubach et al., 2014), sugars (Ingram
141   and Bartels, 1996; Porcel and Ruiz-Lozano, 2004; Rivas-Ubach et al., 2014), and/or other
142   species-specific compounds (Sardans et al., 2011).
143       The metabolome, the chemical phenotype of an organism, is the total set of low molecular
144   weight metabolites (typically <1200 Da) present in an organism at a particular moment (Fiehn,
145   2002). The metabolome, includes thus amino acids, sugars, and nucleotides from primary
146   plant metabolism and many secondary metabolites such as phenolics and terpenes
147   representing the diverse physiological processes in an organism for maintaining internal

                                                        4
148   homeostasis and function. The first functional responses of an organism facing abiotic and
149   biotic stressors are typically at the metabolomic level (Peñuelas and Sardans, 2009).
150   Metabolomics represents thus a powerful tool for ecological studies (ecometabolomics) to
151   identify the main changes in organisms directly associated with metabolism and performance
152   (Sardans et al., 2011). Metabolomics allows us to understand the metabolic variation of
153   organisms under stressful environmental conditions, including the complete set of metabolites
154   and not just single compounds or families of metabolites (Fiehn, 2002; Bundy et al., 2008;
155   Sardans et al., 2011). The study of the metabolomic changes of wild plant species helps to
156   comprehend the mechanisms behind plant physiological responses to natural or experimental
157   stressors. Metabolomic techniques can also assess the plasticity of specific metabolomes and
158   detect and quantify the metabolic biomarkers linked with specific environmental stressors
159   (Bundy et al., 2008; Sardans et al., 2011; Rivas-Ubach et al., 2016a). Ecometabolomics has
160   advanced our understanding of the natural variability and flexibility of the metabolomes of
161   wild organisms under climatic stressors (Gargallo-Garriga et al., 2014; Rivas-Ubach et al.,
162   2014), amongst seasons (Rivas-Ubach et al., 2012), and under attack from folivorous insects
163   (Rivas-Ubach et al., 2016b). Ecometabolomics is thus valuable for exploring the organism-
164   environment interaction by detecting and quantifying the final phenotypic response of an
165   organism to environmental changes.
166       The physiological response to drought of Quercus ilex L., an evergreen sclerophyllous tree
167   species widely distributed in the Mediterranean Basin (Barbero et al., 1992), has been
168   extensively studied (Filella et al., 1998; Nardini et al., 2000; Ogaya and Peñuelas, 2003;
169   Peñuelas et al., 2000; Sala and Tenhunen, 1996). Q. ilex is a keystone species in many
170   Mediterranean ecosystems and is currently expanding its dominance by recolonising
171   abandoned cropland and pastures and by out-competing Mediterranean conifers that are
172   more sensitive to rising temperatures (Carnicer et al., 2013). Drought-induced declines in Q.
173   ilex forests have been reported (Galiano et al., 2012, Camarero et al., 2015), but this tree
174   possesses an array of functional and morphological traits (such as an extensive root system)
175   for surviving periods of drought. Ecometabolomic studies of Q. ilex could thus identify the key
176   metabolites involved in drought tolerance and resistance as well as measure how flexible are
177   the individual metabolomes under stress conditions. We sampled once per season the leaves
178   of differently sized mature Q. ilex trees of the same age from a forest exposed to a moderate
179   experimental drought and analysed the elemental stoichiometries and metabolomes. We
180   discuss three important issues of Q. ilex metabolic responses to environmental variables or
181   factors by multivariate approximations: (i) we hypothesise that the depth of the upper soil
182   layers (DUSL, typically the A + B horizons), where most of the biological activity and water

                                                      5
183   uptake occur, may determine vegetation structure (VS) and overall metabolomic composition;
184   ii) we evaluate how different VSs, with special attention to trees growing in shallower soils,
185   respond to the marked seasonality of the Mediterranean Basin and to experimental drought
186   stress; and iii) we apply the results of this study to illustrate the crucial necessity in
187   ecometabolomic studies of controlling the factors potentially able to produce large
188   metabolomic shifts in plants.
189
190   Material & Methods
191   Study site
192   This study was carried out in a natural Q. ilex forest in the Prades Mountains in southern
193   Catalonia, Spain (41°21’N, 1°2′E). The climate is mesic-Mediterranean, with a marked three-
194   month summer drought. The average annual rainfall is 658 mm, and the average temperature
195   is 12 °C. The forest canopy is dominated by Q. ilex, followed by Phillyrea latifolia and Arbutus
196   unedo, with average trunk basal areas at 50 cm of 20.8, 7.7, and 6.9 m2 ha-1, respectively.
197   Other species adapted to drought conditions are Juniperus oxycedruys, Erica arborea, and
198   Cistus albidus, with sporadic individuals of deciduous species such as Acer monspesulanum and
199   Sorbus torminalis.
200       Tree size in the forest is naturally variable, so VS is also variable. We categorised the VSs by
201   designating the areas dominated by taller trees and higher canopies as H-Forest and the areas
202   dominated by smaller trees and lower canopies as L-Forest. The average trunk diameters at 50
203   cm above the soil (D50s) of the trees of H- and L-Forests were 41.2 and 27.8 cm, respectively.
204
205   Experimental design
206   Eight plots (15  10 m) were established in March 1999 at the same altitude (930 m a.s.l.) on a
207   25% mountain slope facing south-southeast (Ogaya et al., 2003). Four plots were established in
208   each of H- and L-Forest areas to represent the natural variation of the VS of the Q. ilex forest.
209   Two randomly chosen plots for each VS received a drought treatment, and the other two
210   served as control plots. The plots were separated by a minimum of 15 m and were located
211   along the same slope position. The drought treatment consisted of covering approximately
212   30% of the soil surface with 14  1 m PVC strips from the top to the bottom edges of the plots
213   at 0.5-0.8 m above the soil, which partially excluded throughfall. Upslope runoff was
214   intercepted by ditches 0.8-1 m deep dug along the entire top edges of the treatment plots. All
215   water intercepted by the strips and ditches was channelled to the bottom edges of the
216   drought plots.
217

                                                         6
218   Measurement of plot parameters
219   Soil depth was measured by inserting a 1.5-m metallic corer with a pneumatic hammer 0.7-1 m
220   from the trunk of each sampled tree. The depths of the A and B horizons (upper soil layers;
221   DUSL) were measured in situ, and soil samples were collected for calculating the proportion of
222   fine soil (% <2mm). The samples were dried for 72 h at 60 °C and were then sieved through a
223   2-mm mesh, and the fine-soil and stony fractions were weighed. The density of each horizon
224   was calculated. Stone density was considered to be 2.65 g cm-3 (Chepil, 1950). The proportions
225   of fine soil and stones were thus calculated at the tree level. The moisture content of the
226   upper soil was measured during each campaign by time-domain reflectometry (Tektronix
227   1502C, Beaverton, USA) (Gray and Spies, 1995; Zegelin et al., 1989). Three stainless-steel
228   cylindrical rods 25 cm long were vertically installed in the upper 25 cm of the soil at four
229   randomly selected locations in each plot. The time-domain reflectometer was manually
230   attached to the ends of the rods for each measurement.
231       The aboveground biomass (AB) in the plots was estimated by allometric relationships
232   between tree AB and D50. Twelve Q. ilex trees were harvested outside the plots, their
233   circumferences at a height of 50 cm were measured, and their ABs were weighed after drying
234   in an oven to constant weight. We used the calculated allometric relationships to estimate the
235   AB (lnAB = 4.9 + 2.3 lnD50; R2 = 0.98; n = 12; P < 0.001).
236   Moisture content of the upper soil layers and precipitation were monitored every 30 min in all
237   plots (see Ogaya and Peñuelas (2006) for details). Fig. 1a shows the seasonal accumulated
238   annual precipitation in the study area for 1999-2009. Fig. 1b shows the differences of soil
239   moisture amongst the plots and VSs for each season for 2004-2009 (5 years prior to sampling).
240
241   Sampling and processing of leaves
242   We randomly selected five Q. ilex individuals from each plot as study cases. Leaves were
243   sampled once per season February in winter, May in spring, August in summer, and November
244   in autumn. A small sunlit branch was cut from each tree with a pole, and a fraction of the
245   youngest leaves was immediately frozen in liquid nitrogen for the stoichiometric and
246   metabolomic analyses. The youngest leaves were always sampled to standardise the cohort
247   and control any ontogenic shift with season (see Rivas-Ubach et al. (2014) for details). The
248   leaves were collected within a short period between 11:00 and 14:00 to avoid large temporal
249   metabolomic variation (Rivas-Ubach et al., 2013).
250       The leaves were processed as described in detail by Rivas-Ubach et al. (2013). Briefly, the
251   frozen leaves were lyophilised, placed in plastic cans, and stored at -20 °C until ground with a
252   ball mill at 1600 rpm for 6 min (Mikrodismembrator-U, B. Braun Biotech International,

                                                       7
253   Melsungen, Germany) to produce a fine powder for each sample. All samples were then stored
254   in hermetic cans at -80 °C until preparation for metabolite extraction.
255
256   Metabolite extraction for LC-MS
257   Metabolites were extracted as described by t’Kindt et al. (2008) with some modifications. Two
258   sets of 2-mL tubes were labelled (A and B). Set A was used for the extractions, and the extracts
259   were transferred to set B. One hundred milligrams of powder of each foliar sample were
260   introduced into a tube of set A. One millilitre of MeOH/H2O (80:20) was added to each tube.
261   The tubes were vortexed for 15 min and sonicated for 5 min at room temperature (22 °C) and
262   were then centrifuged at 23 000  g for 5 min, and 0.6 mL of supernatant was collected from
263   each tube and transferred to the corresponding tube of set B. This procedure was repeated to
264   perform two extractions of each sample. The tubes of set B were then centrifuged at 23 000  g
265   for 5 min. The supernatants were collected by crystal syringes, filtered with 0.22-µm pore
266   microfilters, and transferred to a labelled set of HPLC vials. The vials were stored at -80 °C until
267   LC-MS analysis.
268
269   Elemental analysis
270   For the C and N analyses, 1.4 mg of powder were transferred to a tin microcapsule. C and N
271   concentrations were determined by elemental analysis using combustion coupled to gas
272   chromatography with a CHNS-O Elemental Analyser (EuroVector, Milan, Italy).
273       P and K concentrations were determined using acid digestion in a microwave at high
274   pressure and temperature (Sardans et al., 2010); 250 mg of powder were weighed in a Teflon
275   tube, and 5 mL of 65% nitric acid and 2 mL of H2O2 were added. The sample was then digested
276   in a MARSXpress microwave (CEM, Mattheus, USA). All digested contents were added to a 50-
277   mL flask and dissolved with Milli-Q water to 50 mL. P and K concentrations were determined
278   by ICP-OES (Optic Emission Spectrometry with Inductively Coupled Plasma) (The Perkin-Elmer
279   Corporation, Norwalk, USA).
280
281   LC-MS analyses
282   LC-MS chromatograms were obtained using a Dionex Ultimate 3000 HPLC system (Thermo
283   Fisher Scientific/Dionex RSLC, Dionex, Waltham, USA) coupled to an LTQ Orbitrap XL high-
284   resolution mass spectrometer (Thermo Fisher Scientific, Waltham, USA) equipped with an HESI
285   II (heated electrospray ionisation) source. Chromatography was performed on a reversed-
286   phase C18 Hypersil gold column (150  2.1 mm, 3 µm particle size; Thermo Scientific,
287   Waltham, USA) at 30 °C. The mobile phases consisted of acetonitrile (A) and 0.1% acetic acid

                                                       8
288   (B). Both mobile phases were filtered and degassed for 10 min in an ultrasonic bath prior to
289   use. The elution gradient began at 10% A (90% B) at a flow rate of 0.3 mL min-1 and was
290   maintained for 5 min, then changed linearly to 10% B (90% A) for the next 15 min. The initial
291   proportions (10% A, 90% B) were gradually recovered over the next 5 min, and the column was
292   then washed and stabilised for 5 min before injection of the next sample. The injection volume
293   of the samples was 5 µL. All samples were injected twice, once with the HESI operating in
294   negative ionisation mode (-H) and once in positive ionisation mode (+H). The Orbitrap mass
295   spectrometer was operated in FTMS (Fourier Transform Mass Spectrometry) full-scan mode
296   with a mass range of 50-1000 m/z and high-mass resolution (60 000). The resolution and
297   sensitivity of the spectrometer were monitored by injecting a caffeine standard after every 10
298   samples, and the resolution was further monitored with lock masses (phthalates). Blank
299   samples were also analysed during the sequence (see Rivas-Ubach et al., 2016b for more
300   details of the instrumentation parameters).
301
302   Processing of LC-MS chromatograms
303   The raw data files from the Orbitrap were processed by MZmine 2.12 (Pluskal et al., 2010).
304   Chromatograms were baseline corrected, deconvoluted, aligned, and autoassigned (see Table
305   A.1 for details). Metabolomic variables were assigned by the exact mass and retention time of
306   our metabolite library performed by the injection of over 200 standards representing several
307   common metabolites of both primary and secondary metabolism (see Table A.2 for
308   assignations). Numerical data sets were exported to CSV and posteriorly filtered; outliers and
309   variables present in fewer than six individuals were removed from the data set. Outlier
310   variables were defined as measurements 3-fold higher than the 3rd quartile or 3-fold lower
311   than the 1st quartile of each cell factor. The areas under the peaks of the deconvoluted
312   chromatograms do not reflect the real concentration as weight of metabolite per weight of the
313   sample but are directly correlated with the concentration of the corresponding variable. The
314   areas are thus suitable for metabolomic comparative analyses, as demonstrated in several
315   metabolomic studies (Gargallo-Garriga et al., 2014; Lee and Fiehn, 2013; Leiss et al., 2013;
316   Mari et al., 2013; Rivas-Ubach et al., 2014, 2016a, 2016b). We use the term concentration to
317   refer to the relative concentration of a metabolite.
318
319   Statistical analyses
320   The data set consisted of three categorical independent variables, season (winter, spring,
321   summer, and autumn), VS (H-Forest and L-Forest), and treatment (control and drought), and
322   2784 dependent continuous variables, ten of which were elemental concentrations and

                                                      9
323   stoichiometric variables (C, N, P, and K concentrations and C/N, N/P, C/P, C/K, N/K, and K/P
324   ratios) and 2774 of which were metabolomic variables, including 35 identified by our plant
325   library.
326       The D50 and DUSL of all trees and the AB of the plots were subjected to one-way ANOVAs
327   with VS as a factor to test for differences between H- and L-Forests (Fig. A.1). Shapiro and
328   Levene’s tests were used on all metabolomic and stoichiometric variables to assess the
329   normality and homogeneity of the variances, respectively. All known variables were normally
330   distributed, and the variances of the groups were homogeneous. Any unidentified
331   metabolomic variable that was not normally distributed or for which the variance of the group
332   was not homogeneous was removed from the data set before statistical analysis to comply
333   with the assumptions of the tests (0.64% of the variables were removed from the data set).
334       The entire data set, including all stoichiometric and metabolomic variables of the Q. ilex
335   leaves (2784 variables), were first subjected to a PERMANOVA analysis using the Bray Curtis
336   distance to test for differences in elemental stoichiometries and metabolomes between
337   seasons, in the experimental drought treatment, and in VS (L-Forest/H-Forest). The number of
338   permutations was set at 10 000 (Table 1). The same data set was also subjected to a principal
339   component analysis (PCA) (Fig. 2) to determine the natural variability amongst samples. The
340   coordinates of cases of the PCA plot were subjected to a one-way ANOVA to identify statistical
341   differences between the analysed groups (See Rivas-Ubach et al. (2013) for details).
342       The Euclidian distances for each Q. ilex individual between control and droughted trees of
343   both H-Forest and L-Forest (HFC vs HFD and LFC vs. LFD, respectively) and between H-Forest
344   and L-Forest controls (HFC. vs. LFC) were calculated using the coordinates of the first 15 PCs of
345   the PCA, which explained the 54.7% of the total variance. The explained variation of each PC
346   was used for the calculation of the Euclidian distance (PC1 (9.7%); PC2 (6.5%); PC3 (6.1%); PC4
347   (5.0%); PC5 (4.5%); PC6 (3.8%); PC7 (3.4%); PC8 (2.8%); PC9 (2.2%); PC10 (2.1%); PC11 (2.0%);
348   PC12 (1.8%); PC13 (1.7%); PC14 (1.6%); PC15 (1.5%)). The distances between groups were then
349   submitted to one-way ANOVAs (Fig. 3).
350       PLS-DAs (partial least squares discriminant analyses) and PCAs were also performed for
351   each season for the elemental and stoichiometric data and the foliar metabolomic fingerprints
352   of the Q. ilex (Figs. 4 and A.2, respectively). The PLS-DAs were performed to clearly identify the
353   variables that could discriminate the various factors studied. The PLS-DAs were represented in
354   bi-plots, and case distribution in multidimensional space was represented by different colours.
355   Individual cases of the PLS-DAs are presented in Fig. A.3.
356       PC2 of the PCA clearly separated the metabolomes of H- and L-Forests (Fig. 2), so we used
357   a structural equation model (SEM) to detect the overall relationships (total, direct, and

                                                     10
358   indirect) of the soil data (DUSL, texture, and moisture) and the experimental drought on D50
359   and the metabolomic variation along PC2 of the PCA of the stoichiometric and metabolomic
360   data from all seasons (Fig. 5; Table 2). We selected the variation along PC2 of the seasonal PCA
361   (Fig. 2) because it clearly separated H- and L-Forests. The data for soil humidity of the SEM
362   corresponded to the seasonal average for 2004-2009. All cases for each season were included
363   in the model, so each Q. ilex tree in each plot and season had a corresponding measure of soil
364   humidity. We incorporated the experimental treatment (categorical variable) into the model
365   by transforming it into a dummy variable (control = 1, drought = 0). Standard errors and the
366   significance levels (P values) of the direct, indirect, and total effects were calculated with
367   bootstrapping (with 1200 repetitions) (Davison et al., 1986). We ran the model with various
368   stipulated relationships and chose the simplest model explaining the most variance of PC2
369   (metabolomic variance) (Fig. 5). The initial model is shown in Fig. A.4, and the total, direct, and
370   indirect effects are shown in Table A.6. Plant metabolomes can vary considerably in response
371   to small environmental shifts, so the SEM model was also performed using the soil humidity of
372   the sampled year (2009-2010), but the overall results were consistent with the seasonal
373   averaged data for 2004-2009. We finally used the 5 year averaged data (2004-2009) of soil
374   moisture for the SEM, because it is a better proxy of tree size than the soil moisture of the
375   sampling year. A heat map of all assigned variables was plotted for all control trees in all
376   seasons to identify the main metabolite shifts amongst seasons and VSs. The relative
377   concentrations for each variable were scaled to the same level before plotting the heat map
378   (Fig. 6).
379       All statistical analyses were performed with R (R Core Team, 2013). The Shapiro tests and
380   ANOVAs were performed with the shapiro.test and aov functions, respectively, in the “R stats”
381   package (R Core Team, 2013). Levene’s test was performed with the levene Test function in the
382   “car” package (Fox and Weisberg, 2011). The PERMANOVA analysis was conducted with the
383   adonis function in the “vegan” package (Oksanen et al., 2013). The PCAs and PLS-DAs were
384   performed by the pca and plsda functions, respectively, of the R “mixOmics” package (Le Cao
385   et al., 2015). The SEM was performed with the sem function of the “sem” package (Fox et al.,
386   2015). The heat map was constructed with the heatmap.2 function in “gplots” package
387   (Warnes et al., 2015). Post-hoc tests were conducted with the HSD.test function in “agricolae”
388   package (de Mendiburu 2015).
389
390   Results
391       D50, DUSL, and AB were higher in H-Forest than in L-Forest (F = 17.7, P < 0.001; F = 11.1, P
392   < 0.01; and F = 60.1, P < 0.0001; respectively) (Fig. A.1). Spring was the wettest season, as

                                                       11
393   expected, followed by winter, autumn, and summer (F = 10.45; P < 0.0001) (Fig. 1a). Autumn
394   was drier than winter from 1999 to 2009, but the difference was not statistically significant (P
395   > 0.05). A factorial ANOVA of soil moisture, including all categorical factors, was significant for
396   season (F = 108.2; P < 0.0001), VS (F = 6.4; P < 0.05), and treatment (F = 13.1; P < 0.001), but
397   none of the interactions were significant. Soil moisture was significantly higher in winter and
398   spring than in summer and autumn (Fig. 1b). The PERMANOVA clearly identified significant
399   differences in the elemental concentrations, stoichiometric ratios, and metabolomes for the Q.
400   ilex leaves between seasons (Pseudo-F = 31.04, P < 0.0001), between VS types (Pseudo-F =
401   3.31, P < 0.05), between experimental treatments (Pseudo-F = 3.98, P < 0.05), and in the
402   VStreatment interaction (Pseudo-F = 9.07, P < 0.001). The seasonVS and seasontreatment
403   interactions were not significant (P > 0.05) (Table 1).
404       The one-way ANOVA of case coordinates of the seasonal PCA identified significant
405   differences amongst seasons along PC1 (F = 10.9, P < 0.0001) and marginal differences along
406   PC2 (F = 2.27, P = 0.08) (marginal differences are not indicated in the graph), but the case plot
407   did not clearly cluster cases for each season (Fig. 2b). The cases, however, were better
408   separated between the two VSs in the annual PCA in both PCs, but especially in PC2 (F = 7.48, P
409   < 0.01 for PC1 and F = 351, P < 0.0001 for PC2) (Fig. 2c). The experimental drought treatment
410   produced significant separation along PC1 when excluding VS (F = 15.5, P < 0.001). The
411   significance of the drought treatment, however, was extended to PC2 within each VS (H-
412   Forest: F = 7.75, P < 0.01 for PC1 and F = 5.89, P < 0.05 for PC2; L-Forest: F = 58.9, P < 0.0001
413   for PC1 and F = 6.71, P < 0.05 for PC2) (Fig. 2c). The metabolomes of the leaves shifted most
414   strongly across seasons, but VS and the experimental drought also had significant effects
415   (Table 1).
416       One-way ANOVAs of the metabolomic distances (Euclidian distances) calculated with the
417   first 15 PCs of the PCA showed that the HFC vs. HFD distance in each of the seasons was
418   smaller than the HFC vs. LFC distance. The LFC vs. LFD and HFC vs. LFC distances, however, did
419   not differ for summer, autumn, or winter, and the LFC vs. LFD distance was even larger in
420   spring (Fig. 3). The LFC vs. LFD distance was largest in winter and spring, the wetter seasons.
421       The factorial ANOVAs of all known variables, excluding the effects of seasonality (seasons
422   taken together because the treatmentseason and VSseason interactions were not significant
423   in the PERMANOVA, see Table 1), identified several variables with a significant VStreatment
424   effect, suggesting that the drought treatment affected each VS differently (Table A.3). All
425   elemental and stoichiometric variables, except C concentrations, and metabolites such as
426   pentoses, valine, arginine, malic acid, epigallocatechin, homoorientin, chlorogenic acid, -


                                                      12
427   humulene, caryophyllene, and pyridoxine responded significantly to the VStreatment
428   interaction (P < 0.05). The factorial ANOVAs of all individual variables for the VStreatment
429   interaction for each season separately did not have many common responses with VS and
430   treatment for specific variables amongst the seasons (Table A.4). The PLS-DA, however, clearly
431   indicated some common responses of VS and treatment with the seasons (Fig. 4). The changes
432   amongst VS and the treatments were not significant for all seasons, but L-Forest under the
433   drought treatment generally had higher concentrations of P, N, -humulene, caryophyllene,
434   disaccharides, pentoses, hexoses, leucine, succininc acid, lactic acid, chlorogenic acid, and
435   catechin and higher N/K and C/K ratios (Fig. 4). The PLS-DAs, as expected, identified larger
436   metabolomic differences along components 1 and 2 of the case plot between the control and
437   droughted plants in L-Forest than in H-Forest in all seasons (Figs 4 and A.2).
438       The SEM indicated that the proportion of fine soil was not correlated with the moisture
439   content of the upper soil, and DUSL did not influence the soil texture; soil texture was thus not
440   included in the final model. All included variables (drought treatment, upper soil moisture,
441   DUSL, and D50) had significant direct or indirect effects on the metabolomes of the leaves
442   along PC2, explaining 38.4% of its total variation. D50 explained most of the variation, followed
443   by the indirect effect of DUSL (Fig. 5; Table 2). DUSL and the effect of the experimental drought
444   explained 48.8% of the variance of the D50 of the Q. ilex trees, with DUSL the most significant
445   factor (R2 = 0.59, P < 0.00001). Of the variables included in the model, only the drought
446   treatment had a significant effect on upper soil moisture, explaining 3.5% of the total variance
447   (Fig. 5; Table 2).
448       The factorial ANOVAs of all known variables for the control trees, with season and VS as
449   categorical factors, indicated that several of the variables shifted significantly amongst seasons
450   and VSs (Fig. 6). Biomarkers of drought and oxidative stress, i.e. epigallocatechin, epicatechin,
451   catechin, gallic acid, quinic acid, kaempferol, rhamnetin, proline, and quercitin, did not differ
452   significantly between VSs. Concentrations of quinic acid were higher in summer. The N:P ratio
453   was generally lowest and the concentrations of some amino acids were generally higher in
454   trees in spring, although not significantly (Fig. 6).
455
456   Discussion
457   Metabolomic variation, VS, and drought.
458       The drought treatment significantly affected the metabolomes and stoichiometries of the
459   natural populations of Q. ilex (Table 1), as also reported for other plant species (Gargallo-
460   Garriga et al., 2014; Rivas-Ubach et al., 2012; J. Sardans et al., 2012c; Sardans et al., 2011 and
461   citations therein; Urano et al., 2009). The effects of the drought treatment on the foliar

                                                        13
462   metabolomes and stoichiometries differed in the L- and H-Forests, as shown by the
463   PERMANOVA (Table 1) and the case plot of the seasonal PCA, with the two VSs clearly
464   separated along PC2 (Fig. 2c). Interestingly, the metabolomic distance in all seasons was
465   smaller between the control and droughted trees in H-Forest (HFC vs HFD) than between the
466   control trees of H-Forest and L-Forest (HFC vs LFC) (Fig. 3), indicating that tree metabolism was
467   affected more by VS than by the experimental drought treatment. The distance between the
468   control and droughted trees, however, was significantly larger in all seasons in L-Forest than in
469   H-Forest and this distance was larger in winter and spring, the wetter seasons, than in summer
470   and autumn (Fig. 3). Moreover, case clustering in the PLS-DAs identified larger differences
471   between the control and droughted trees in L-Forest than in H-Forest (Fig. 4). These results
472   demonstrated a higher metabolomic sensitivity to drought of trees in shallower soils (L-Forest),
473   supporting the importance of VS to the response to drought. In fact, shallower soil should be
474   an important factor in VS evolution from continuous to patchy vegetation cover under
475   increasingly arid conditions in desertified areas of the Mediterranean (Sardans and Peñuelas,
476   2014). In this scenario, areas with deeper soils would be more likely to remain covered. The
477   large difference in the tree metabolomes between the two VSs exposed to the same
478   experimental stress demonstrated the necessity of including VS as an important factor when
479   designing experiments, especially metabolomic studies, in natural ecosystems. Most plants
480   require well-balanced soil humidity to maintain physiological homeostasis, fitness, and growth.
481   The SEM, however, indicated that the moisture content of the upper soil did not explain the
482   variation in D50 of the plots (Fig. 5), in contrast to the results of previous studies (Huxman et
483   al., 2004; Knapp and Smith, 2001; Orwig and Abrams, 1997). Other Mediterranean trees also
484   depend on water stored in deep soil layers or in rock fractures to cope with summer drought
485   (Barbeta et al., 2015, Voltas et al., 2015). The upper soil layers are typically very dry during
486   summer (Fig. 1), so growth is determined more by the availability of deep water pools that
487   may allow trees to reduce the period of growth cessation than by the amount of moisture in
488   the upper layers (Lempereur et al., 2015). The effect of the drought treatment on soil moisture
489   was noticeable in all seasons, including winter, when deep-water pools are replenished
490   (Brooks et al., 2010). The effect the treatment on growth and mortality rates has been
491   dampened over time by the larger reductions in density in the drought plots (Barbeta et al.,
492   2013), but extremely dry summers still affect the drought plots more negatively (Barbeta et al.,
493   2015). The negative effect of the drought treatment on D50 may thus be mediated by the
494   depletion of deep-water pools caused by a chronically lower replenishment during cold
495   seasons (Barbeta et al., 2015).



                                                       14
496       Interestingly, the moisture content of the upper soil was nevertheless significantly
497   correlated with the metabolomic variation of Q. ilex L-Forest along PC2 (Fig. 5), which
498   separated not only the VSs but also the experimental drought treatment within each VS (Fig.
499   2). The treatment, however, was not correlated with PC2, probably due the larger separation
500   of VSs than the drought treatment along PC2. The significant relationship between soil
501   moisture and PC2 potentially illustrates the effects of water availability on the metabolomes of
502   plants, as has been reported in other metabolomic studies (Gargallo-Garriga et al., 2014;
503   Griesser et al., 2015; Rivas-Ubach et al., 2012, 2014; Sardans et al., 2011 and citations therein;
504   Sun et al., 2014; Zhang et al., 2014). DUSL was not correlated with the moisture content of the
505   upper soil, as expected, but was highly correlated with D50, which together with the effects of
506   the experimental drought explained 48.6% of the D50 variance (Fig. 5). Hydraulic lift has been
507   described in Q. ilex (David et al., 2007), but the SEM suggested that tree size (D50) and
508   therefore VS were more dependent on the changes in DUSL than on upper soil moisture,
509   indicating that upper soil moisture alone could not exclusively account for the VSs. The large
510   metabolomic distance between the control and droughted trees (Figs. 2 and 3) and the
511   significant VStreatment interaction (Table 1), however, suggest that moisture content of the
512   upper soil still played an important role in determining the physiology and homeostasis of the
513   trees (Huxman et al., 2004; Rosenzweig, 1968; Sala et al., 1988), especially those under the
514   effects of drought and in shallower soils (L-Forest).
515       Seasonality had a large effect on the Q. ilex metabolomes, in accordance with previous
516   metabolomic studies (Rivas-Ubach et al., 2012, 2014), but the effect varied significantly
517   between VSs (Fig. 6). The smaller distance between the control and droughted trees of L-
518   Forest in summer and autumn, the driest seasons (Fig. 1), suggest that the control trees
519   experienced some level of natural drought stress in those seasons (Figs. 1 and 3). This seasonal
520   trend, however, was not significant in H-Forest. The significant shift with season in
521   metabolomic distance between the control and droughted trees in L-Forest, largest in winter
522   and spring and smallest in summer and autumn (Fig. 3), indicated that the responses of Q. ilex
523   to drought in shallower soils were tightly linked with seasonality. This variable response of VS
524   to seasonality also suggests that Q. ilex trees in shallower soils are more sensitive to the
525   typically large natural environmental changes in Mediterranean ecosystems. Drivers of global
526   change such as drought can lead to long-term changes in VS by reducing the growth of trees
527   (Barber et al., 2000; Bréda et al., 2006), which may consequently negatively impact soil fertility
528   (Gallardo et al., 2000) and enzymatic activity (Garcia et al., 2002) and may increase runoff
529   (Garcia-Estringana et al., 2010; Goberna et al., 2007; Johnson-Maynard et al., 2002), thereby
530   reducing soil depth. Shallower soils can generally have negative feedbacks on plants (Ruiz-

                                                      15
531   Sinoga et al., 2011; Sadaka and Ponge, 2003), which may result in a higher vulnerability to
532   increases in the intensity, frequency, and duration of drought forecasted for the coming
533   decades, especially in the Mediterranean Basin (IPCC, 2007).
534
535
536   Foliar chemical changes, drought, and VS.
537   Foliar K concentrations were similar in the two VSs and in the droughted and control trees (Fig.
538   4; Table A.4). K concentrations, however, interestingly varied significantly amongst seasons,
539   with the highest concentrations in summer, the driest season (Figs. 2 and 3). Several plant
540   species allocate large amounts of K to leaves during the driest season, especially in
541   Mediterranean evergreen and dry tropical forests (Milla et al., 2005; Rivas-Ubach et al., 2012;
542   Sardans and Peñuelas 2015). K is involved in several plant mechanisms that mitigate drought
543   stress (Sardans and Peñuelas, 2015), for example the maintenance of cellular turgor and
544   osmotic pressure (Ashraf et al., 2002; Levi et al., 2011), the control of transpiration and water
545   conductance (Arquero et al., 2006; Benlloch-González et al., 2010; Harvey and van den
546   Driessche, 1999), the control of stomata (Benlloch-González et al., 2010; Talbott and Zeiger,
547   1996), and the maintenance of transmembrane potentials (Su et al., 2001; Waraich et al.,
548   2011). K can even act as an osmolyte (Babita et al., 2010; Levi et al., 2011) and can thus
549   improve the capacity to retain water (Nandwal et al., 1998). K concentrations were not higher
550   in the droughted trees but were higher in all trees in the driest season, as reported in other
551   studies (Sardans and Peñuelas 2015 and citations therein), suggesting that increases in foliar K
552   concentration are a programmed phenological response of plants to cope with the driest
553   season.
554       Oxidative stress is expected to increase in plants under conditions of drought, especially in
555   water-limited ecosystems such as those in the Mediterranean Basin (Price et al., 1989; Dat et
556   al., 2000; Munné-Bosch & Peñuelas, 2004; Peñuelas et al., 2004). The concentrations of
557   antioxidants has been reported to increase in water-stressed plants for coping with oxidative
558   stress, so these compounds are excellent biomarkers of drought stress (Farooq et al., 2009;
559   Reddy et al., 2004; Wang et al., 2003). Their role as electron donors and the ability to alter
560   peroxidation kinetics are the main properties that provide their strong antioxidant activity
561   (Rice-Evans et al., 1997). The concentrations of most of the antioxidants identified in our study
562   (epigallocatechin, epicatechin, catechin, gallic acid, quinic acid, kaempferol, rhamnetin, and
563   quercitin) did not differ significantly between VSs (Figs. 4 and 6). Additionally, the
564   concentration of proline, a well-known multifunctional amino acid that typically acts as
565   osmoregulator under drought conditions (Szabados and Savouré, 2010), was not significantly

                                                      16
566   higher in L-Forest than in H-Forest (Figs. 4 and 6; Tables A.3 and A.4), suggesting that L-Forest
567   trees, even growing in shallower soils, were not necessarily more water-stressed than H-Forest
568   trees. The levels of most antioxidant compounds were not significantly higher in summer (Fig.
569   6), but flavonoid concentrations have been reported to be generally higher in summer in H-
570   Forest trees (Rivas-Ubach et al., 2014). The concentration of quinic acid was higher in summer,
571   in accordance with previous studies (Rivas-Ubach et al., 2012, 2014), especially in L-Forest
572   trees (Fig. 6). Several flavonoids are synthesised from quinic acid, tyrosine, and phenylalanine
573   by the shikimic acid pathway (Draths et al., 1999; Harborne 1988), suggesting a potential
574   activation of this metabolic route in summer to cope with oxidative stress. The metabolism of
575   amino acids is complex; each amino acid is involved in several metabolic pathways, which
576   complicates the interpretation of changes in their concentrations. The concentrations of
577   several amino acids, however, were higher in spring, especially when compared to summer
578   (Fig. 6).
579       Spring is the principal growing season in Mediterranean ecosystems, and the higher
580   concentrations of several amino acids in spring may have been associated with higher growth
581   rates (Rivas-Ubach et al., 2012). N:P ratios were also generally lowest in spring, and the
582   tendency of lower N:P ratios in H-Forest trees than L-Forest trees potentially indicates the
583   higher growth rate in spring in H-Forest than L-Forest according to the growth rate hypothesis
584   (Elser et al., 1996). The concentrations of malic, pyruvic, and citric acids also tended to be
585   higher in H-Forest trees, which may also account for the higher growth than in L-Forest trees
586   (Fig. 6). The increase in carbohydrates in droughted plants is a common strategy to cope with
587   drought stress. Carbohydrates can act as osmolytes, improving the hydric potential and thus
588   maintaining cellular turgor (Ingram and Bartels, 1996; Leprince et al., 1993; Porcel and Ruiz-
589   Lozano, 2004; Rivas-Ubach et al., 2014). The concentrations of carbohydrates were higher in
590   our droughted L-Forest trees than in the other plots, including the H-Forest droughted plants
591   (Figs. 2 and 4), which potentially indicates a higher drought stress in the droughted L-Forest
592   trees.
593       We have reported some clues indicating drought stress in droughted Q. ilex trees, but
594   plants can use various metabolic pathways to cope with drought depending on the intensity of
595   the stress. Such responses are also typically species-specific, so some plant species may use
596   proportionally higher concentrations of metabolites than others for osmoprotection (Leprince
597   et al., 1993; Ingram and Bartels, 1996). The acumulation of carbohydrates in leaves, however,
598   may also be due to the impairment of growth as drought progresses. The C-loading of trees is
599   higher during periods of zero growth, such as the summer cessation of growth in Holm oak
600   forests (Lemepreur et al., 2015), because source activity (photosynthesis) exceeds sink activity

                                                      17
601   (growth) (Estiarte and Peñuelas 1999; Körner, 2003; Peñuelas and Estiarte 1998). The
602   accumulation of carbohydrates can consequently be an acclimative mechanism, but it may also
603   be a passive process indicating the impairment of sink activities at the individual level (Estiarte
604   and Peñuelas 1999; Peñuelas and Estiarte 1998).
605       The synthesis of terpenes is typically associated with attacks by folivores and pathogens
606   and is one of the main induced defensive chemical mechanisms in plants (Achotegui-Castells et
607   al., 2013; Huang et al., 2012; Köpke et al., 2010; Mumm and Hilker, 2006; Pare and Tumlinson,
608   1997). L-Forest droughted trees tended to have higher foliar concentrations of caryophyllene
609   and -humulene (Fig. 4), suggesting a higher herbivory pressure in L- than H-Forest trees
610   during drought. The synthesis of caryophyllene, an indirect defensive compound that attracts
611   predators and parasitoids of the herbivore (Köllner et al., 2008; Rasmann et al., 2005), is
612   increased under folivory (Gouinguené et al., 2001; Rivas-Ubach et al., 2016b). Rivas-Ubach et
613   al. (2014) suggested that droughted H-Forest Q. ilex trees had significantly higher levels of
614   folivory associated with the foliar concentrations of carbohydrates and antioxidants due
615   drought stress. The higher foliar terpene concentrations in the droughted L-Forest trees found
616   in our study, in addition to the increased drought stress, provide clues for the potentially
617   higher susceptibility of these trees to herbivory and/or pathogenic stress (Gaylord et al., 2013;
618   Rivas-Ubach et al., 2014; Rouault et al., 2006; Desprez-Loustau et al., 2006).
619
620   Metabolomics in natural ecosystems.
621   This study has demonstrated that metabolomic techniques are excellent tools for identifying
622   metabolic shifts in plants and other organisms under conditions of stress (Sardans et al., 2011;
623   Shulaev et al., 2008). These techniques are sufficiently sensitive to detect small shifts in the
624   metabolomes of plants over seasons (Rivas-Ubach et al., 2012); under experimental drought
625   treatments (Urano et al., 2009) and herbivorous attack (Rivas-Ubach et al., 2016a, 2016b); with
626   salinity (Sanchez et al., 2008), warming (Gargallo-Garriga et al., 2015; Rivas-Ubach et al., 2012),
627   nutrient toxicity (Navascués et al., 2012), and ultraviolet light (Broeckling et al., 2005); and
628   amongst other stressors or environmental gradients (Sardans et al., 2011 and citations
629   therein). Plant metabolomes can vary widely amongst individuals of the same species, even
630   under the same environmental conditions (Gargallo-Garriga et al., 2015, 2014; Rivas-Ubach et
631   al., 2014, 2012), which is indicative of their high plasticity. The metabolomic shifts detected in
632   response to each of the factors in this study (season, treatment, and VS) are likely to have
633   different directions, which prevents a clear clustering of cases from the same groups in the
634   PCA and complicates the interpretation of the results for each of the factors (Figs. 2 and 6). It is
635   also clear in the PERMANOVA showing the significance of the VStreatment interaction

                                                       18
636   suggesting that the effects of the drought treatment vary between VSs (Table 1). The large
637   metabolomic variation between H- and L-Forests in our analyses (see Figs. 2, 3, 4 and 6)
638   compelled us to include VS as another significant independent factor, which is not often
639   included in ecophysiological field studies. The high sensitivity of metabolomics for detecting
640   significant differences of slight metabolic shifts requires an awareness of any environmental
641   factor that could potentially affect the composition of the metabolomes.
642       Our results have thus demonstrated the necessity of identifying and including in
643   metabolomic studies any factor that may produce large metabolomic variation amongst
644   individuals within the same group level of the primary factors studied (seasons and drought
645   treatment in our study). Any uncontrolled and/or unknown environmental factor causing
646   substantial shifts in the metabolomes could lead to inconclusive or even incorrect
647   interpretations of the results for the primary factors. All ecometabolomic studies should thus
648   pay special attention to including any factor in the experimental design that may significantly
649   affect the metabolomes of plants.
650
651
652   Conclusions
653   · The metabolome of Q. ilex trees was highly dependent on the vegetation structure defined by
654   tree size (H-Forest, L-Forest).
655   · The depth rather than the moisture content of the upper soil layers was correlated with
656   vegetation structure, suggesting that shallow soils may not be able to sustain large Q. ilex
657   trees.
658   · The effects of drought on the metabolomes of Q. ilex trees were stronger in trees in
659   shallower soils (L-Forest). The depth of the upper soil layers is thus a potential factor
660   determining the future of tree populations under drought conditions and possible further
661   desertification, especially in Mediterranean ecosystems.
662   · Potassium concentrations were higher in summer, as in other studies, but were not higher in
663   trees growing in shallower soils nor in droughted trees, supporting our hypothesis that the
664   increase in K concentrations is a programmed phenological response of plants to cope with dry
665   seasons.
666   · Metabolomics is a very sensitive technique for detecting shifts in metabolomes amongst
667   individuals and for improving our understanding of the relationships of metabolomes with
668   environmental variables. This high sensitivity indicates the necessity of identifying and
669   analysing any component able to strongly affect the metabolomes of individuals within the
670   same group.

                                                      19
671
672
673
674
675
676
677
678
679   Acknowledgements                                                                                Formatat: Cap

680   The authors thank Sara Férez, Gemma Montalban, and Laia Mateu-Castell for their field and
681   laboratory support. ARU appreciates the financial support of the research fellowship (JAE)
682   from the CSIC, and AB acknowledges an FPI predoctoral fellowship from the Ministry of
683   Economy and Competitiveness of Spain. This research was supported by the European
684   Research Council Synergy grant SyG-2013-610028 IMBALANCE-P, the Spanish Government
685   projects CGL2013-48074-P and OAPN 022/2008 (PROPINOL), and the Catalan Government
686   project SGR 2014-274. A portion of the research was performed using EMSL, a DOE Office of
687   Science User Facility sponsored by the Office of Biological and Environmental Research at the
688   Pacific Northwest National Laboratory. MO and OU were supported by the grant projects
689   M200871201 (AS CR), CZ.1.07/2.3.00/20.0246 (MSMT), and LO1415 (MSMT).
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705

                                                    20
706
707
708
709
710
711
712
713
714   Reference list


715   Achotegui-Castells, A., Llusia, J., Hódar, J.A., Peñuelas, J., 2013. Needle terpene concentrations
716       and emissions of two coexisting subspecies of Scots pine attacked by the pine
717       processionary moth (Thaumetopoea pityocampa). Acta Physiol. Plant. 35, 3047–3058.

718   Allen, C.D., Breshears, D.D., 1998. Drought-induced shift of a forest-woodland ecotone: Rapid
719        landscape response to climate variation. Proc. Natl. Acad. Sci. 95, 14839–14842.

720   Allen, C.D., Macalady, A.K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M.,
721        Kitzberger, T., Rigling, A., Breshears, D.D., Hogg, E.H. (Ted), Gonzalez, P., Fensham, R.,
722        Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G., Running, S.W., Semerci, A., Cobb,
723        N., 2010. A global overview of drought and heat-induced tree mortality reveals emerging
724        climate change risks for forests. For. Ecol. Manage. 259, 660–684.

725   Arquero, O., Barranco, D., Benlloch, M., 2006. Potassium Starvation Increases Stomatal
726       Conductance in Olive Trees. HortScience 41, 433–436.
727
728   Ashraf, M., Ashfaq, M., Ashraf, M.Y., 2002. Effects of Increased Supply of Potassium on Growth
729        and Nutrient Content in Pearl Millet under Water Stress. Biol. Plant. 45, 141–144.
730
731   Aschmann, H., 1973. Distribution and Peculiarity of Mediterranean Ecosystems. Mediterranean
732       Type Ecosystems 7, 11-19.
733
734   Babita, M., Maheswari, M., Rao, L.M., Shanker, A.K., Rao, D.G., 2010. Osmotic adjustment,
735        drought tolerance and yield in castor (Ricinus communis L.) hybrids. Environ. Exp. Bot. 69,
736        243–249.

737   Barber, V.A., Juday, G.P., Finney, B.P., 2000. Reduced growth of Alaskan white spruce in the
738        twentieth century from temperature-induced drought stress. Nature 405, 668–73.
739
740   Barbero, M., Loisel, R., Quézel, P., 1992. Biogeography, ecology and history of Mediterranean
741        Quercus ilex ecosystems. Vegetatio 99, 19-34.
742
743   Barbeta, A., Ogaya, R., Peñuelas, J., 2013. Dampening effects of long-term experimental
744        drought on growth and mortality rates of a Holm oak forest. Glob. Chang. Biol. 19, 3133–
745        44.
746


                                                      21
747   Barbeta, A., Mejía-Chang, M., Ogaya, R., Voltas, J., Dawson, T.E., Peñuelas, J., 2015. The
748        combined effects of a long-term experimental drought and an extreme drought on the
749        use of plant-water sources in a Mediterranean forest. Glob. Chang. Biol. 12, 1213-1225.
750
751   Bastida, F., Barberá, G.G., García, C., Hernández, T., 2008. Influence of orientation, vegetation
752        and season on soil microbial and biochemical characteristics under semiarid conditions.
753        Appl. Soil Ecol. 38, 62–70.

754   Benlloch-González, M., Romera, J., Cristescu, S., Harren, F., Fournier, J.M., Benlloch, M., 2010.
755        K+ starvation inhibits water-stress-induced stomatal closure via ethylene synthesis in
756        sunflower plants. J. Exp. Bot. 61, 1139–45.
757
758   Bertram, H.C., Weisbjerg, M.R., Jensen, C.S., Pedersen, M.G., Didion, T., Petersen, B.O., Duus,
759        J.Ø., Larsen, M.K., Nielsen, J.H., 2010. Seasonal changes in the metabolic fingerprint of 21
760        grass and legume cultivars studied by nuclear magnetic resonance-based metabolomics.
761        J. Agric. Food Chem. 58, 4336–41.

762   Bigler, C., Bräker, O.U., Bugmann, H., Dobbertin, M., Rigling, A., 2006. Drought as an Inciting
763         Mortality Factor in Scots Pine Stands of the Valais, Switzerland. Ecosystems 9, 330–343.

764   Bowman, W.D., Theodose, T.A., Schardt, J.C., Conant, R.T., 1993. Constraints of nutrient
765      availability on primary production in two alpine tundra communities. Ecology 74, 2085-
766      2097.

767   Bréda, N., Huc, R., Granier, A., Dreyer, E., 2006. Temperate forest trees and stands under
768        severe drought: a review of ecophysiological responses, adaptation processes and long-
769        term consequences. Ann. For. Sci. 63, 625–644.
770
771   Broeckling, C.D., Huhman, D. V, Farag, M.A., Smith, J.T., May, G.D., Mendes, P., Dixon, R.A.,
772        Sumner, L.W., 2005. Metabolic profiling of Medicago truncatula cell cultures reveals the
773        effects of biotic and abiotic elicitors on metabolism. J. Exp. Bot. 56, 323–36.
774
775   Brooks, J.R., Barnard, H.R., Coulombe, R., McDonnell, J.J., 2010. Ecohydrologic separation of
776        water between trees and streams in a Mediterranean climate. Nat. Geosc. 3, 100-104.

777   Bundy, J.G., Davey, M.P., Viant, M.R., 2008. Environmental metabolomics: a critical review and
778       future perspectives. Metabolomics 5, 3–21.

779   Cakmak, I., Engels, C., 1999. Role of mineral nutrients in photosynthesis and yield formation,
780       in: Rengel, Z. (Ed.), Mineral Nutrition of Crops: Mechanisms and Implications. The
781       Haworth Press, New York, NY, pp. 141–168.
782
783   Caldwell, M.M., Dawson, T.E., Richards, J.H., 1998. Hydraulic lift: consequences of water efflux
784        from the roots of plants. Oecologia 113, 151–161.
785
786   Caldwell, M.M., Richards, J.H., 1989. Hydraulic lift: water efflux from upper roots improves
787        effectiveness of water uptake by deep roots. Oecologia 79, 1–5.
788
789   Camarero, J.J., Sangüesa-Barreda, G. and Vergarechea, M. 2015. Prior height, growth, and
790       wood anatomy differently predispose to drought-induced dieback in two Mediterranean
791       oak species. Ann. For. Sci.. DOI 10.1007/s13595-015-0523-4.


                                                      22
792
793   Canadell, J., Jackson, R.B., Ehleringer, J.B., Mooney, H.A., Sala, O.E., Schulze, E.-D., 1996.
794       Maximum rooting depth of vegetation types at the global scale. Oecologia 108, 583–595.
795
796   Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sánchez, G., Peñuelas, J., 2011. Widespread
797        crown condition decline, food web disruption, and amplified tree mortality with
798        increased climate change-type drought. Proc. Natl. Acad. Sci. U. S. A. 108, 1474–8.
799
800   Carnicer, J., Barbeta, A., Sperlich, D., Coll, M., Peñuelas, J., 2013. Constrasting trait syndromes
801        in angiosperms and conifers are associated with different responses of tree growth to
802        temperature on a large scale. Front. Plant Sci. 4:409.
803
804   Chepil, W., 1950. Methods of estimating apparent density of discrete soil grains and
805       aggregates. Soil Sci. 70, 351–362.

806   David, T.S., Henriques, M.O., Kurz-Besson, C., Nunes, J., Valente, F., Vaz, M., Pereira, J.S.,
807        Siegwolf, R., Chaves, M.M., Gazarini, L.C., David, J.S., 2007. Water-use strategies in two
808        co-occurring Mediterranean evergreen oaks: surviving the summer drought. Tree Physiol.
809        27, 793–803.

810   Davison, A.C., Hinkley, D. V., Schechtman, E., 1986. Efficient Bootstrap Simulation. Biometrika
811        73, 555.
812
813   de Mendiburu, F., 2015. Agricolae: Statistical Procedures for Agricultural Research. R package
814       version 1.2-3.
815
816   Desprez-Loustau, M.-L., Marçais, B., Nageleisen, L.-M., Piou, D., Vannini, A., 2006. Interactive
817       effects of drought and pathogens in forest trees. Ann. For. Sci. 63, 597–612.
818
819   Draths, K.M., Knop, R.D., Frost, J.W., 1999. Shikimic acid and wuinic acid: Replacing isolation
820        from plant sources with recombinant microbial biocatalysis. J. Am. Chem. Soc. 121, 1603-
821        1604.
822
823   Elser, J.J., Dobberfuhl, D.R., Mackay, N.A., Schampel, J.H., 1996. Organism size, life history and
824         N:P stoichiometry. BioScience 46, 674-684.

825   Epstein, H.E., Lauenroth, W.K., Burke, I.C., 1997. Effects of temperature and soil texture on
826        anpp in the u.s. great plains. Ecology 78, 2628–2631.
827
828   Estiarte, M., Peñuelas, J., 1999. Excess carbon: the relationship with phenotypical plasticity in
829         storage and defense functions of plants. Orsis 14, 159-203.

830   Fahrig, L., 2003. Effects of habitat fragmentation on biodiversity. Annu. Rev. Ecol. Evol. Syst.
831        34, 487–515.

832   Falasca, A., Melck, D., Paris, D., Saviano, G., Motta, A., Iorizzi, M., 2013. Seasonal changes in
833        the metabolic fingerprint of Juniperus communis L. berry extracts by 1H NMR-based
834        metabolomics. Metabolomics 10, 165-174.




                                                       23
835   Farahani, H.J., Peterson, G.A., Westfall, D.G., Sherrod, L.A., Ahuja, L.R., 1998. Soil Water
836        Storage in Dryland Cropping Systems: The Significance of Cropping Intensification. Soil
837        Sci. Soc. Am. J. 62, 984.

838   Farooq, M., Wahid, A., Kobayashi, N., Fujita, D., Basra, S.M.A., 2009. Plant drought stress:
839        effects, mechanisms and management. Agron. Sustain. Dev. 29, 185–212.

840   Fernandez-Illescas, C.P., Porporato, A., Laio, F., Rodriguez-Iturbe, I., 2001. The ecohydrological
841        role of soil texture in a water-limited ecosystem. Water Resour. Res. 37, 2863–2872.
842
843   Fiehn, O., 2002. Metabolomics - the link between genotypes and phenotypes. Plant Mol. Biol.
844        48, 155–171.
845
846   Fierer, N., Schimel, J.P., Holden, P.A., 2003. Variations in microbial community composition
847        through two soil depth profiles. Soil Biol. Biochem. 35, 167–176.

848   Filella, I., Llusià, J., Piñol, J., Peñuelas, J., 1998. Leaf gas exchange and fluorescence of Phillyrea
849          latifolia, Pistacia lentiscus and Quercus ilex saplings in severe drought and high
850          temperature conditions. Environ. Exp. Bot. 39, 213–220.

851   Fox, J., Nie, Z., Byrnes, J., 2015. SEM: Structural Equation Models. R Package. version 3.1-6.
852
853   Fox, J., Weisberg, S., 2011. An {R} Companion to Applied Regression. R package.
854
855   Fuhlendorf, S.D., Smeins, F.E., 1998. The influence of soil depth on plant species response to
856        grazing within a semi-arid savanna. Plant Ecol. 138, 89–96.
857
858   Galiano, L., Martínez-Vilalta, J., Lloret, F., 2010. Drought-Induced Multifactor Decline of Scots
859        Pine in the Pyrenees and Potential Vegetation Change by the Expansion of Co-occurring
860        Oak Species. Ecosystems 13, 978–991.
861
862   Galiano, L., Martínez-Vilalta, J., Lloret, F., 2011. Carbon reserves and canopy defoliation
863        determine the recovery of Scots pine 4 yr after a drought episode. New Phytol. 190, 750–
864        9.
865   Galiano, L., Martínez-Vilalta, J., Sabaté, S., Lloret, F., 2012. Determinants of drought effects on
866        crown condition and their relationship with depletion of carbon reserves in a
867        Mediterranean holm oak forest. Tree Physiol. 32, 478–89.
868
869   Gallardo, A., Rodríguez-Saucedo, J.J., Covelo, F., Fernández-Alés, R., 2000. Soil nitrogen
870        heterogeneity in a Dehesa ecosystem. Plant Soil 222, 71–82.
871
872   Garcia, C., Hernandez, T., Roldan, A., Martin, A., 2002. Effect of plant cover decline on chemical
873        and microbiological parameters under Mediterranean climate. Soil Biol. Biochem. 34,
874        635–642.
875
876   Garcia-Estringana, P., Alonso-Blázquez, N., Marques, M.J., Bienes, R., Alegre, J., 2010. Direct
877        and indirect effects of Mediterranean vegetation on runoff and soil loss. Eur. J. Soil Sci.
878        61, 174–185.
879
880   Gargallo-Garriga, A., Sardans, J., Pérez-Trujillo, M., Oravec, M., Urban, O., Jentsch, A., Kreyling,
881        J., Beierkuhnlein, C., Parella, T., Peñuelas, J., 2015. Warming differentially influences the


                                                        24
882        effects of drought on stoichiometry and metabolomics in shoots and roots. New Phytol.
883        207, 591–603.
884
885   Gargallo-Garriga, A., Sardans, J., Pérez-Trujillo, M., Rivas-Ubach, A., Oravec, M., Vecerova, K.,
886        Urban, O., Jentsch, A., Kreyling, J., Beierkuhnlein, C., 2014. Opposite metabolic responses
887        of shoots and roots to drought. Sci. Rep. 4.
888
889   Gaylord, M.L., Kolb, T.E., Pockman, W.T., Plaut, J.A., Yepez, E.A., Macalady, A.K., Pangle, R.E.,
890        McDowell, N.G., 2013. Drought predisposes piñon-juniper woodlands to insect attacks
891        and mortality. New Phytol. 198, 567–78.
892
893   Goberna, M., Sánchez, J., Pascual, J.A., García, C., 2007. Pinus halepensis Mill. plantations did
894       not restore organic carbon, microbial biomass and activity levels in a semi-arid
895       Mediterranean soil. Appl. Soil Ecol. 36, 107–115.

896   Gouinguené, S., Degen, T., Turlings, T.C.J., 2001. Variability in herbivore-induced odour
897        emissions among maize cultivars and their wild ancestors (teosinte). Chemoecology 11,
898        9–16.

899   Gray, A.N., Spies, T.A., 1995. Water content measurement in forest soils and decayed wood
900        using time domain reflectometry. Can. J. For. Res. 25, 376–385.
901
902   Griesser, M., Weingart, G., Schoedl-Hummel, K., Neumann, N., Becker, M., Varmuza, K.,
903        Liebner, F., Schuhmacher, R., Forneck, A., 2015. Severe drought stress is affecting
904        selected primary metabolites, polyphenols, and volatile metabolites in grapevine leaves
905        (Vitis vinifera cv. Pinot noir). Plant Physiol. Biochem. 88, 17–26.

906   Hanson, P.J., Weltzin, J.F., 2000. Drought disturbance from climate change: response of United
907       States forests. Sci. Total Environ. 262, 205–220.
908
909   Harborne, J.B., 1988. The Flavonoids: Advances in Research Since 1986, Chapman & Hall, New
910       York, pp 23–54.
911
912   Harrison, R.D., 2001. Drought and the consequences of El Niño in Borneo: a case study of figs.
913        Popul. Ecol. 43, 63–75.
914
915   Harvey, H.P., van den Driessche, R., 1999. Nitrogen and potassium effects on xylem cavitation
916        and water-use efficiency in poplars. Tree Physiol. 19, 943–950.

917   Huang, M., Sanchez-Moreiras, A.M., Abel, C., Sohrabi, R., Lee, S., Gershenzon, J., Tholl, D.,
918       2012. The major volatile organic compound emitted from Arabidopsis thaliana flowers,
919       the sesquiterpene (E)-β-caryophyllene, is a defense against a bacterial pathogen. New
920       Phytol. 193, 997–1008.
921
922   Huxman, T.E., Smith, M.D., Fay, P.A., Knapp, A.K., Shaw, M.R., Loik, M.E., Smith, S.D., Tissue,
923       D.T., Zak, J.C., Weltzin, J.F., Pockman, W.T., Sala, O.E., Haddad, B.M., Harte, J., Koch,
924       G.W., Schwinning, S., Small, E.E., Williams, D.G., 2004. Convergence across biomes to a
925       common rain-use efficiency. Nature 429, 651–4.
926
927   Ingram, J., Bartels, D., 1996. The molecular basis of dehydration tolerance in plants. Annu. Rev.
928        Plant Biol. 47, 377–403.

                                                      25
929
930   IPCC. 2007. Core Writing Team, Pachauri RK, Reisinger A. Fourth Assessment Report: Climate
931        change 2007 (AR4) Contribution of Working Groups I, II and III to the Fourth Assessment
932        Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: IPCC.
933
934   Jobbagy, E.G., Jackson, R.B., 2000. The Vertical Distribution of Soil Organic Carbon and Its
935        Relation to Climate and Vegetation. Ecol. Appl. 10, 423.
936
937   Johnson-Maynard, J., Graham, R., Wu, L., Shouse, P., 2002. Modification of soil structural and
938        hydraulic properties after 50 years of imposed chaparral and pine vegetation. Geoderma
939        110, 227–240.
940
941   Kinnell, P.I.A., Cummings, D., 1993. Soil/slope gradient interactions in erosion by rain-impacted
942        flow. Trans. ASAE. 36, 381-387.
943
944   Knapp, A.K., Smith, M.D., 2001. Variation among biomes in temporal dynamics of aboveground
945       primary production. Science 291, 481–4.

946   Köllner, T.G., Held, M., Lenk, C., Hiltpold, I., Turlings, T.C.J., Gershenzon, J., Degenhardt, J.,
947        2008. A maize (E)-beta-caryophyllene synthase implicated in indirect defense responses
948        against herbivores is not expressed in most American maize varieties. Plant Cell 20, 482–
949        94.
950
951   Köpke, D., Beyaert, I., Gershenzon, J., Hilker, M., Schmidt, A., 2010. Species-specific responses
952       of pine sesquiterpene synthases to sawfly oviposition. Phytochemistry 71, 909–17.

953   Köpke, D., Beyaert, I., Gershenzon, J., Hilker, M., Schmidt, A., 2010. Species-specific responses
954       of pine sesquiterpene synthases to sawfly oviposition. Phytochemistry 71, 909–17.
955
956   Körner, C. 2003. Carbon limitation in trees. J. of Ecol. 91, 4–17.

957   Kuske, C.R., Ticknor, L.O., Busch, J.D., Gehring, C.A., Whitham, T.G., 2003. The pinyon
958        rhizosphere, plant stress, and herbivory affect the abundance of microbial decomposers
959        in soils. Microb. Ecol. 45, 340–52.

960   Le Cao, L-A., Gonzalez, I., Dejean, S., 2015. mixOmics: Omics Data Integration Project. R
961        package version 5.2.0.
962
963   Lee, D.Y., Fiehn, O., 2013. Metabolomic response of Chlamydomonas reinhardtii to the
964        inhibition of target of rapamycin (TOR) by rapamycin. J. Microbiol. Biotechnol. 23, 923–
965        31.
966
967   Leiss, K.A., Cristofori, G., van Steenis, R., Verpoorte, R., Klinkhamer, P.G.L., 2013. An eco-
968         metabolomic study of host plant resistance to Western flower thrips in cultivated,
969         biofortified and wild carrots. Phytochemistry 93, 63–70.
970
971   Lempereur, M., Martin-StPaul, N.K., Damesin, C., Joffre, R., Ourcival J.M., Rocheteau, A.,
972       Rambal, S., 2015. Growth duration is a better predictor of stem increment than carbon
973       supply in a Mediterranean oak forest: implications for assessing forest productivity under
974       climate change. New Phytol. 207, 579-590.



                                                       26
 975   Leprince, O., Hendry, G.A.F., McKersie, B.D., 1993. The mechanisms of desiccation tolerance in
 976        developing seeds. Seed Sci. Res. 3, 231–246.
 977
 978   Levi, A., Paterson, A.H., Cakmak, I., Saranga, Y., 2011. Metabolite and mineral analyses of
 979         cotton near-isogenic lines introgressed with QTLs for productivity and drought-related
 980         traits. Physiol. Plant. 141, 265–75.

 981   Lloret, F., Siscart, D., Dalmases, C., 2004. Canopy recovery after drought dieback in holm-oak
 982        Mediterranean forests of Catalonia (NE Spain). Glob. Chang. Biol. 10, 2092–2099.
 983
 984   Lloret, F., Escudero, A., Iriondo, J.M., Martínez-Vilalta, J., Valladares, F., 2012. Extreme climatic
 985        events and vegetation: the role of stabilizing processes. Glob. Chang. Biol. 18, 797-805.
 986
 987   Mari, A., Lyon, D., Fragner, L., Montoro, P., Piacente, S., Wienkoop, S., Egelhofer, V.,
 988        Weckwerth, W., 2013. Phytochemical composition of Potentilla anserina L. analyzed by
 989        an integrative GC-MS and LC-MS metabolomics platform. Metabolomics 9, 599–607.
 990
 991   McDowell, N., Pockman, W.T., Allen, C.D., Breshears, D.D., Cobb, N., Kolb, T., Plaut, J., Sperry,
 992       J., West, A., Williams, D.G., Yepez, E.A., 2008. Mechanisms of plant survival and mortality
 993       during drought: why do some plants survive while others succumb to drought? New
 994       Phytol. 178, 719–39.
 995
 996   Milla, R., Castro-Díez, P., Maestro-Martínez, M., Montserrat-Martí, G., 2005. Relationships
 997         between phenology and the remobilization of nitrogen, phosphorus and potassium in
 998         branches of eight Mediterranean evergreens. New Phytol. 168, 167–78.
 999
1000   Montheith, J., 1981. Soil Water and Nitrogen in Mediterranean-type Environments. London.

1001   Mueller, R.C., Scudder, C.M., Porter, M.E., Talbot Trotter, R., Gehring, C.A., Whitham, T.G.,
1002       2005. Differential tree mortality in response to severe drought: evidence for long-term
1003       vegetation shifts. J. Ecol. 93, 1085–1093.
1004
1005   Mumm, R., Hilker, M., 2006. Direct and indirect chemical defence of pine against folivorous
1006      insects. Trends Plant Sci. 11, 351–8.
1007
1008   Nandwal, A.S., Hooda, A., Datta, D., 1998. Effect of Substrate Moisture and Potassium on
1009       Water Relations and C, N and K Distribution in Vigna Radiata. Biol. Plant. 41, 149–153.
1010
1011   Nardini, A., Salleo, S., Gullo, M.A. Lo, Pitt, F., 2000. Different responses to drought and freeze
1012        stress of Quercus ilex L. growing along a latitudinal gradient. Plant Ecol. 148, 139–147.

1013   Navascués, J., Pérez-Rontomé, C., Sánchez, D.H., Staudinger, C., Wienkoop, S., Rellán-Álvarez,
1014      R., Becana, M., 2012. Oxidative stress is a consequence, not a cause, of aluminum toxicity
1015      in the forage legume Lotus corniculatus. New Phytol. 193, 625–36.
1016
1017   Nepstad, D.C., de Carvalho, C.R., Davidson, E.A., Jipp, P.H., Lefebvre, P.A., Negreiros, G.H., da
1018       Silva, E.D., Stone, T.A., Trumbore, S.E., Vieira, S., 1994. The role of deep roots in the
1019       hydrological and carbon cycles of Amazonian forests and pastures. Nature 372, 666–669.

1020   Ogaya, R., Peñuelas, J., 2003. Comparative field study of Quercus ilex and Phillyrea latifolia:
1021       photosynthetic response to experimental drought conditions. Environ. Exp. Bot. 50, 137–


                                                        27
1022        148.

1023   Ogaya, R., Peñuelas, J., 2006. Tree growth, mortality, and above-ground biomass accumulation
1024       in a holm oak forest under a five-year experimental field drought. Plant Ecol. 189, 291–
1025       299.
1026    gaya, R., e uelas, ., ar nez-Vilalta, J., Mangirón, M., 2003. Effect of drought on diameter
1027       increment of Quercus ilex, Phillyrea latifolia, and Arbutus unedo in a holm oak forest of
1028       NE Spain. For. Ecol. Manage. 180, 175–184.
1029
1030   Oksanen, J., Guillaume-Blanchet, F., Kindt, R., Legendre, ., inchin, ., ’Hara, R., Simpson,
1031       G., Solymos, P., Stevens, M., Wagner, H., 2013. vegan: Community Ecology Package.
1032
1033   Orwig, D.A., Abrams, M.D., 1997. Variation in radial growth responses to drought among
1034       species, site, and canopy strata. Trees 11, 474.
1035
1036   Pare, P.W., Tumlinson, J.H., 1997. De Novo Biosynthesis of Volatiles Induced by Insect
1037        Herbivory in Cotton Plants. Plant Physiol. 114, 1161–1167.
1038
1039   Peñuelas, J., Esitarte., 1998. Can elevated CO2 affect secondary metabolism and ecosystem
1040       function? Trends. Ecol. Evol. 13, 20-24.
1041
1042   Peñuelas, J., Filella, I., 2003. Deuterium labelling of roots provides evidence of deep water
1043       access and hydraulic lift by Pinus nigra in a Mediterranean forest of NE Spain. Environ.
1044       Exp. Bot. 49, 201–208.

1045   Peñuelas, J., Filella, I., Lloret, F., Piñol, J., Siscart, D., 2000. Effects of a Severe Drought on
1046       Water and Nitrogen Use by Quercus ilex and Phyllyrea latifolia. Biol. Plant. 43, 47–53.

1047   Peñuelas, J., Sardans, J., 2009. Ecological metabolomics. Chem. Ecol. 25, 305–309.
1048
1049   Pluskal, T., Castillo, S., Villar-Briones, A., rešič, ., 2010. MZmine 2: modular framework for
1050        processing, visualizing, and analyzing mass spectrometry-based molecular profile data.
1051        BMC Bioinformatics 11, 395.

1052   Poorter, L., 1999. Growth responses of 15 rain-forest tree species to a light gradient: the
1053        relative importance of morphological and physiological traits. Funct. Ecol. 13, 396–410.

1054   Porcel, R., Ruiz-Lozano, J.M., 2004. Arbuscular mycorrhizal influence on leaf water potential,
1055        solute accumulation, and oxidative stress in soybean plants subjected to drought stress. J.
1056        Exp. Bot. 55, 1743–50.
1057
1058   Poyatos, R., Aguadé, D., Galiano, L., Mencuccini, M., Martínez-Vilalta, J., 2013. Drought-
1059        induced defoliation and long periods of near-zero gas exchange play a key role in
1060        accentuating metabolic decline of Scots pine. New Phytol. 200, 388–401.
1061
1062   Pregitzer, K.S., Laskowski, M.J., Burton, A.J., Lessard, V.C., Zak, D.R., 1998. Variation in sugar
1063        maple root respiration with root diameter and soil depth. Tree Physiol. 18, 665–670.
1064
1065   Prieto, I., Armas, C., Pugnaire, F.I., 2012. Water release through plant roots: new insights into
1066        its consequences at the plant and ecosystem level. New Phytol. 193, 830–41.
1067


                                                         28
1068   R Core Team, 2013. R: A language and environment for statistical computing.

1069   Rasmann, S., Köllner, T.G., Degenhardt, J., Hiltpold, I., Toepfer, S., Kuhlmann, U., Gershenzon,
1070       J., Turlings, T.C.J., 2005. Recruitment of entomopathogenic nematodes by insect-
1071       damaged maize roots. Nature 434, 732–7.

1072   Rebetez, M., Dobbertin, M., 2004. Climate change may already threaten Scots pine stands in
1073       the Swiss Alps. Theor. Appl. Climatol. 79, 1–9.

1074   Reddy, A.R., Chaitanya, K.V., Vivekanandan, M., 2004. Drought-induced responses of
1075       photosynthesis and antioxidant metabolism in higher plants. J. Plant Physiol. 161, 1189–
1076       1202.
1077
1078   Rice-Evans, C., Miller, N., Paganga, G., 1997. Antioxidant properties of phenolic compounds.
1079        Trends Plant Sci. 2, 152–159.

1080   Rivas-Ubach, A., Gargallo-Garriga, A., Sardans, J., Oravec, M., Mateu-Castell, L., Pérez-Trujillo,
1081        M., Parella, T., Ogaya, R., Urban, O., Peñuelas, J., 2014. Drought enhances folivory by
1082        shifting foliar metabolomes in Quercus ilex trees. New Phytol. 202, 874–885.
1083
1084   Rivas-Ubach, A., Pérez-Trujillo, M., Sardans, J., Gargallo-Garriga, A., Parella, T., Peñuelas, J.,
1085        2013. Ecometabolomics: optimized NMR-based method. Methods Ecol. Evol. 4, 464–473.
1086
1087   Rivas-Ubach, A., Hódar, J.A., Sardans, J., Kyle, J., Kim, Y-M., Oravec, M., Urban, O., Guenther,
1088        A., Peñuelas, J., 2016a. Are the metabolomic responses to folivory of closely related plant
1089        species linked to macroevolutionary and plant-folivory coevolutionary processes? Ecol.
1090        Evol. Doi: 10.1002/ece3.2206
1091
1092   Rivas-Ubach, A., Sardans, J., Hódar J.A., Garcia-Porta, J., Guenther, A., Oravec, M., Urban, O.,
1093        Peñuelas, J., 2016b. Similar local but different systemic metabolomic responses of closely
1094        related pine subspecies to folivory by caterpillars of the processionary moth. Plant Biol.
1095        18, 484-494.
1096
1097   Rivas-Ubach, A., Sardans, J., Pérez-Trujillo, M., Estiarte, M., Peñuelas, J., 2012. Strong
1098        relationship between elemental stoichiometry and metabolome in plants. Proc. Natl.
1099        Acad. Sci. 109, 4181–4186.
1100
1101   Rosenzweig, M.L., 1968. Net Primary Productivity of Terrestrial Communities: Prediction from
1102       Climatological Data. Am. Nat. 102, 67.
1103
1104   Rouault, G., Candau, J.-N., Lieutier, F., Nageleisen, L.-M., Martin, J.-C., Warzée, N., 2006. Effects
1105       of drought and heat on forest insect populations in relation to the 2003 drought in
1106       Western Europe. Ann. For. Sci. 63, 613–624.
1107
1108   Ruiz-Sinoga, J.D., Gabarrón Galeote, M.A., Martinez Murillo, J.F., Garcia Marín, R., 2011.
1109        Vegetation strategies for soil water consumption along a pluviometric gradient in
1110        southern Spain. CATENA 84, 12–20.
1111
1112   Sadaka, N., Ponge, J.F., 2003. Climatic effects on soil trophic networks and the resulting humus
1113        profiles in holm oak (Quercus rotundifolia) forests in the High Atlas of Morocco as
1114        revealed by correspondence analysis. Eur. J. Soil Sci. 54, 767–777.


                                                        29
1115   Sala, A., Tenhunen, J.D., 1996. Simulations of canopy net photosynthesis and transpiration in
1116         Quercus ilex L. under the influence of seasonal drought. Agric. For. Meteorol. 78, 203–
1117         222.

1118   Sala, O.E., Parton, W.J., Joyce, L.A., Lauenroth, W.K., 1988. Primary Production of the Central
1119         Grassland Region of the United States. Ecology 69, 40-45.
1120
1121   Sanchez, D.H., Siahpoosh, M.R., Roessner, U., Udvardi, M., Kopka, J., 2008. Plant metabolomics
1122        reveals conserved and divergent metabolic responses to salinity. Physiol. Plant. 132, 209–
1123        19.
1124
1125   Sardans, J., Montes, F., Peñuelas, J., 2010. Determination of As, Cd, Cu, Hg and Pb in biological
1126        samples by modern electrothermal atomic absorption spectrometry. Spectrochim. Acta
1127        Part B At. Spectrosc. 65, 97–112.
1128
1129   Sardans, J., Peñuelas, J., 2014. Hydraulic redistribution by plants and nutrient stoichiometry:
1130        Shifts under global change. Ecohydrology 7, 1–20.
1131
1132   Sardans, J., Peñuelas, J., 2015. Potassium: a neglected nutrient in global change. Glob. Ecol.
1133        Biogeogr. 24, 261-275.
1134
1135   Sardans, J., Peñuelas, J., Coll, M., Vayreda, J., Rivas-Ubach, A., 2012a. Stoichiometry of
1136        potassium is largely determined by water availability and growth in Catalonian forests.
1137        Funct. Ecol. 26, 1077–1089.
1138
1139   Sardans, J., Peñuelas, J., Rivas-Ubach, A., 2011. Ecological metabolomics: overview of current
1140        developments and future challenges. Chemoecology 21, 191–225.
1141
1142   Sardans, J., Rivas-Ubach, A., Peñuelas, J., 2012c. The C:N:P stoichiometry of organisms and
1143        ecosystems in a changing world: A review and perspectives. Perspect. Plant Ecol. Evol.
1144        Syst. 14, 33–47.

1145   Sardans, J., Rivas-Ubach, A., Peñuelas, J., 2012b. The elemental stoichiometry of aquatic and
1146        terrestrial ecosystems and its relationships with organismic lifestyle and ecosystem
1147        structure and function: a review and perspectives. Biogeochemistry 111, 1–39.
1148

1149   Schulze, E.-D., Mooney, H.A., Sala, O.E., Jobbagy, E., Buchmann, N., Bauer, G., Canadell, J.,
1150        Jackson, R.B., Loreti, J., Oesterheld, M., Ehleringer, J.R., 1996. Rooting depth, water
1151        availability, and vegetation cover along an aridity gradient in Patagonia. Oecologia 108,
1152        503–511.

1153   Schwinning, S., Sala, O.E., Loik, M.E., Ehleringer, J.R., 2004. Thresholds, memory, and
1154       seasonality: understanding pulse dynamics in arid/semi-arid ecosystems. Oecologia 141,
1155       191–3.

1156   Shulaev, V., Cortes, D., Miller, G., Mittler, R., 2008. Metabolomics for plant stress response.
1157        Physiol. Plant. 132, 199–208.

1158   Su, H., Golldack, D., Katsuhara, M., Zhao, C., Bohnert, H.J., 2001. Expression and stress-
1159        dependent induction of potassium channel transcripts in the common ice plant. Plant


                                                       30
1160        Physiol. 125, 604–14.

1161   Sun, C., Gao, X., Fu, J., Zhou, J., Wu, X., 2014. Metabolic response of maize (Zea mays L.) plants
1162        to combined drought and salt stress. Plant Soil 388, 99–117.

1163   Szabados, L., Savouré, A., 2010. Proline: a multifunctional amino acid. Trends Plant Sci. 15, 89–
1164        97.

1165   t’Kindt, R., De Veylder, L., Storme, ., Deforce, D., Van Bocxlaer, ., 2008. LC-MS metabolic
1166         profiling of Arabidopsis thaliana plant leaves and cell cultures: optimization of pre-LC-MS
1167         procedure parameters. J. Chromatogr. B. Analyt. Technol. Biomed. Life Sci. 871, 37–43.

1168   Talbott, L.D., Zeiger, E., 1996. Central Roles for Potassium and Sucrose in Guard-Cell
1169        Osmoregulation. Plant Physiol. 111, 1051–1057.

1170   Tymms, M.J., Gaff, D.F., 1979. Proline Accumulation during Water Stress in Resurrection Plants.
1171       J. Exp. Bot. 30, 165–168.
1172
1173   Urano, K., Maruyama, K., Ogata, Y., Morishita, Y., Takeda, M., Sakurai, N., Suzuki, H., Saito, K.,
1174       Shibata, D., Kobayashi, M., Yamaguchi-Shinozaki, K., Shinozaki, K., 2009. Characterization
1175       of the ABA-regulated global responses to dehydration in Arabidopsis by metabolomics.
1176       Plant J. 57, 1065–78.
1177
1178   Voltas, J., Lucabaugh, D., Chambel, M.R., Ferrio, J.P., 2015. Intraspecific variation in the use of
1179        water sources by the circum-Mediterranean conifer Pinus halepensis. 208, 1031-1041.
1180
1181   Wan, C., Xu, W., Sosebee, R.E., Machado, S., Archer, T., 2000. Hydraulic lift in drought-tolerant
1182       and -susceptible maize hybrids. Plant Soil 219, 117–126.

1183   Wang, W., Vinocur, B., Altman, A., 2003. Plant responses to drought, salinity and extreme
1184       temperatures: towards genetic engineering for stress tolerance. Planta 218, 1–14.
1185
1186   Waraich, E.A., Ahmad, R., Ashraf, M., 2011. Role of Mineral Nutrition in Alleviation of Drought
1187       Stress in Plants. Aust. J. Crop Sci. 5, 764.

1188   Wardle, D.A., Bardgett, R.D., Klironomos, J.N., Setälä, H., van der Putten, W.H., Wall, D.H.,
1189       2004. Ecological linkages between aboveground and belowground biota. Science 304,
1190       1629–33.
1191
1192   Warnes, G.R., Bolker, B., Bonebakker, L., Gentleman, R., Huber W., Liaw, A., Lumley, T.,
1193       Maechler, M., Magnusson, A., Moeller, S., Schwartz, M., Venables, B., 2015.
1194       gplots: Various R Programming Tools for Plotting Data. R package version 2.17.0.

1195   Yamada, M., Morishita, H., Urano, K., Shiozaki, N., Yamaguchi-Shinozaki, K., Shinozaki, K.,
1196       Yoshiba, Y., 2005. Effects of free proline accumulation in petunias under drought stress. J.
1197       Exp. Bot. 56, 1975–81.
1198
1199   Zegelin, S.J., White, I., Jenkins, D.R., 1989. Improved field probes for soil water content and
1200        electrical conductivity measurement using time domain reflectometry. Water Resour.
1201        Res. 25, 2367–2376.
1202

                                                        31
1203   Zhang, J.-Y., Cruz DE Carvalho, M.H., Torres-Jerez, I., Kang, Y., Allen, S.N., Huhman, D. V, Tang,
1204       Y., Murray, J., Sumner, L.W., Udvardi, M.K., 2014. Global reprogramming of transcription
1205       and metabolism in Medicago truncatula during progressive drought and after rewatering.
1206       Plant. Cell Environ. 37, 2553–76.

1207
1208
1209
1210
1211
1212
1213
1214   Table 1. Full factorial PERMANOVA model (excluding the triple interaction) of the complete
1215   metabolomic and stoichiometric data set: season, vegetation structure (VS), drought
1216   treatment, seasonVS, seasondrought treatment, and VSdrought treatment. Bold type
1217   indicate significant effects.
1218

                                       Degrees
                                       of        Sum of      Mean of
                                       freedom   squares     squares       Pseudo-F    P

       Season                          3         4.06        1.35          31.0        0.00001

       Vegetation structure (VS)       1         0.14        0.14          3.31        0.044

       Drought treatment               1         0.17        0.17          3.98        0.025

       SeasonVS                       3         0.12        0.04          0.90        0.47

       SeasonTreatment                3         0.17        0.06          1.33        0.24

       VSTreatment                    1         0.40        0.40          9.07        0.0005

       Residuals                       147       6.40        0.04          0.56

       Total                           159       11.5        1.00

1219
1220
1221
1222
1223


                                                        32
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235   Table 2. Correlation coefficients (R2) and the corresponding P values for the total, direct, and
1236   indirect effects of the simplified structural equation model (SEM) with tree diameter (D50),
1237   moisture content of the upper soil, and metabolomic variation (PC2) as endogenous variables
1238   and drought treatment, moisture content of the upper soil, depth of the upper soil layers
1239   (DUSL), and D50 as exogenous variables. The model is described in Figure 5.
1240

       TOTAL EFFECTS
                                                   Upper soil
                      Drought treatment                                 DUSL               D50
                                                   moisture

       D50            0.39 (P < 0.001)                                  0.64 (P < 0.001)

       Upper soil
                      0.19 (P < 0.001)
       moisture

       PC2            0.25 (P < 0.001)             0.24 (P < 0.001)     0.34 (P < 0.001)   0.53 (P < 0.001)



       DIRECT EFFECTS
                                                   Upper soil
                       Drought treatment                                DUSL               D50
                                                   moisture

       D50             0.39 (P < 0.001)                                 0.64 (P < 0.001)

       Upper soil
                       0.19 (P < 0.001)
       moisture

       PC2                                         0.24 (P < 0.001)                        0.53 (P < 0.001)



       INDIRECT EFFECTS
                                                   Upper soil
                       Drought treatment                                DUSL               D50
                                                   moisture

                                                       33
       D50

       Upper soil
       moisture

       PC2             0.25 (P < 0.001)                                  0.34 (P < 0.001)

1241
1242
1243
1244
1245
1246
1247   Figure Captions.
1248
1249   Figure 1. Box plots indicating the medians and distributions of a) seasonal precipitation (mm)
1250   for 1999-2009 and b) seasonal soil moisture (v/v.) in the upper soil layers for each control and
1251   drought plot from H-Forest and L-Forest for 2004-2009. Black and white circles indicate the
1252   medians for the control and drought plots, respectively. H- and L-Forest boxes are white and
1253   grey, respectively. Different letters denote statistical differences after Tukey’s post-hoc tests (P
1254   < 0.05).
1255




1256
1257
1258
1259
1260
1261
1262

                                                       34
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275   Figure 2. Principal component (PC) 1 vs. PC2 of the principal component analysis (PCA) of the
1276   metabolomic and stoichiometric variables over four seasons for the leaves of Q. ilex. (a)
1277   Carbon (C), nitrogen (N), phosphorus (P), and potassium (K) ratios are shown in red.
1278   Metabolomic families are indicated by different colours: blue, sugars; green, amino acids;
1279   cyan, nucleotides; orange, organic acids of the tricarboxylic acid cycle; yellow, phenolics; dark
1280   red, other metabolites; violet, terpenes. Unassigned metabolomic variables are not
1281   represented in the graph. Metabolite abbreviations: Disacch (disaccharides), Hex (hexoses),
1282   Pent (pentoses), Ala (alanine), Arg (arginine), Asp (aspartic acid), Glu (glutamic acid), Leu
1283   (leucine), Phe (phenylalanine), Pro (proline), Trp (tryptophan), Tyr (tyrosine), Val (valine), Cit.ac
1284   (citric acid), Mal.ac (malic acid), Succ.ac (succinic acid), Pyr (pyruvate), Lac.ac (lactic acid), Cat
1285   (catechin), Chlo.ac (chlorogenic acid), Hom (homoorientin), Epi (epicatechin), Epigallo
1286   (epigallocatechin), Gall.ac (gallic acid), Kaemp (kaempferol), Lut (luteolin), Rham (rhamnetin),
1287   aHum (-humulene), Cary (caryophyllene), Quer (quercetin), Qui.ac (quinic acid), and Pyrid
1288   (pyridoxine). (b) Case plot representing the cases by season (blue for winter, green for spring,
1289   red for summer, and orange for autumn). The coloured arrows indicate the means for PC1 of
1290   the corresponding season. (c) Case plot representing the cases by vegetation structure and
1291   experimental drought treatment. Black arrows indicate the means for the PC1 and PC2 of the
1292   corresponding vegetation structure (L-Forest or H-Forest). The coloured arrows indicate the
1293   means for PC1 and PC2 of the corresponding treatment (control or drought) in each vegetation
1294   structure. Different letters by the arrows indicate significant differences between seasons,
1295   vegetation structures, or experimental treatments (P < 0.05).
1296


                                                         35
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310



       36
1311   Figure 3. Metabolomic distance (based on Euclidian distances) for each season of the paired
1312   groups of samples H-Forest-Control vs H-Forest-Drought (HFC vs HFC, squares), L-Forest-
1313   Control vs L-Forest-Drought (LFC vs LFD, circles), and H-Forest-Control vs L-Forest-Control (HFC
1314   vs LFC, triangles). Different letters denote statistical significance amongst all distances after
1315   Tukey’s post-hoc tests (P<0.05).
1316
1317
                                                                                                           Formatat: Centrada




1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333

                                                        37
1334   Figure 4. Bi-plots of component 2 vs. component 1 of the partial least squares discriminant
1335   analysis of the metabolomic and stoichiometric variables for the leaves of Q. Ilex for (a) winter,
1336   (b) spring, (c) summer, and (d) autumn. The variables in the plots are the same as in Figure 2
1337   and have the same codes. The areas of distribution of the cases are coloured by vegetation
1338   structure and experimental treatment: light green for H-forest controls, dark green for H-
1339   Forest droughted trees, orange for L-Forest controls, and red for L-Forest droughted trees. The
1340   original representations of the cases are shown in Figure A.3.
1341




1342
1343
1344
1345
1346


                                                      38
1347   Figure 5. Simplified structural equation model including the moisture content of the upper soil,
1348   tree diameter (D50), and metabolomic variation (PC2) as endogenous variables and depth of
1349   the upper soil layers (DUSL) and experimental drought treatment as exogenous variables. The
1350   model coefficient (R2) and the corresponding P value for each relationship are shown between
1351   the variables. Total, direct, and indirect effects are shown in Table 2.
1352
                                                                                                          Formatat: Centrada




1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366



                                                       39
1367   Figure 6. P values of factorial ANOVAs for each known variable with season and vegetation
1368   structure (VS) as categorical factors (a). Significant and marginally significant P values are
1369   shown in red and blue, respectively. Heat map (hierarchical variable relationship) of the known
1370   variables for each season (winter, spring, summer, and autumn) and vegetation structure (H-
1371   Forest (H-F) and L-Forest (L-F)) (b). The colours represent the relative metabolite concentration
1372   amongst each of the groups. Red represents the highest concentration, and greater colour
1373   differences for each variable indicate larger differences amongst the groups. Different letters
1374   indicate statistically significance after HSD post-hoc test (P < 0.05).
1375




1376

                                                        40
1377   Appendix A
1378   Topsoil depth substantially influences the tree metabolomes of Mediterranean forests.
1379   Rivas-Ubach et al., 2015
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409




                                                    41
1410   Table A.1. LC-MS chromatogram processing. Chromatograms were obtained by LC-MS and
1411   processed by MZmine 2.12. The following table summarizes the processes and parameters
1412   applied to the Q. ilex foliar metabolomic chromatograms.

                                                       (+H) Chromatograms           (-H) Chromatograms
       1      Baseline correction
                           Chromatogram type                       TIC                          TIC
                                     MS level                        1                            1
                                    Smoothing                      10E7                         10E7
                                   Asymmetry                       0.001                        0.001
       2      Mass detection (exact Mass)
                                    Noise level                   5 × 105                      4 × 105
       3      Chromatogram builder
                         Minimum time span                         0.05                         0.05
                           Minimum height                         25000                        25000
                              m/z tolerance                       0.002                        0.002
       4      Smoothing
                                       Filter width                 5                            5
       5      Chromatogram deconvolution
              (local minimum search)
                     Chromatographic threshold                     65%                          65%
                   Search minimum in RT range                       0.1                          0.1
                                          (min)
                       Minimum relative height                     5.0%                         5.0%
                      Minimum absolute height                     30000                        30000
                Minimum ratio of peak top/edge                       2                            2
                            Peak duration range                   0.0-2.0                      0.0-2.0
       6      Chromatogram alignment (join
              alignment)
                               m/z tolerance                       0.001                        0.001
                             Weight for m/z                         80                           80
                               RT tolerance                        0.15                          0.2
                              Weight for RT                         20                           20
       7      Gap filling (Peak Finder)
                             Intensity tolerance                   20%                          20%
                                   m/z tolerance                   0.001                        0.001
                        Retention time tolerance                    0.1                          0.1
                                     RT correction                marked                       marked
       8      Filtering
                       Minimum peaks in a row                       8                            8


       9      Metabolite Assignation
                                m/z tolerance                     0.0005                       0.0005
                                 RT tolerance                      0.25                         0.25
              Ions excluded from database                           <75                         <85
                                                                   83.05                      119.035
                                                                  102.05                      223.082
                                                                  114.09                      391.196
                                                                  227.17                       159.25
                                                                  607.29                      186.186
                                                           Between 0.0 and 1 min       Between 0.0 and 1.1 min
                                                          Between 28.5 and 30 min      Between 27.0 and 30 min
1413   RT, retention time; m/z, mass to charge ratio

                                                                   42
1414   Table A.2. Metabolite assignment by LC-MS. The assignment of the metabolites was based on
1415   the standards. The following table summarizes the retention time (RT) and mass to charge
1416   ratio (m/z) of the assigned metabolites in both positive and negative ionization modes.
1417
                   Mode        Compound                       RT (min)           m/z
                   Negative    Catechin                          3.44         289.0718
                   Negative    Chlorogenic acid                  3.11         353.0873
                   Negative    Citric acid                       1.77         191.0196
                   Negative    Deoxyhexose                       1.42         163.0618
                   Negative    Disaccharides                     1.43          341.108
                   Negative    Epicatechin                    4.93 - 5.2      289.0713
                   Negative    Epigallocatechin              1.54 - 2.64      305.0667
                   Negative    Gallic acid                   1.55 - 1.83      169.0147
                   Negative    Hexose                            1.44          179.056
                   Negative    Homoorientin                      9.45         447.0923
                   Negative    Kaempferol                       14.82         285.0404
                   Negative    Lactic acid                   1.52; 1.75        89.0245
                   Negative    Malic acid                    1.51; 1.78       133.0143
                   Negative    Pentose                           1.43         149.0456
                   Negative    Pyruvate                          1.65          87.0089
                   Negative    Quercetin                        13.72         301.0355
                   Negative    Quinic acid                       1.47          191.056
                   Negative    Rhamnetin                        15.98         315.0509
                   Negative    Sodium salicylate                10.51         137.0245
                   Negative    Succinic acid                 1.74; 1.78       117.0194
                   Positive    Adenine                       1.42; 1.77       136.0614
                   Positive    Adenosine                     1.49; 1.75       268.1038
                   Positive    α-Humulene                       20.27         205.1949
                   Positive    Alanine                           1.43           90.054
                   Positive    Arginine                          1.34          175.119
                   Positive    Aspartic acid                      1.5          134.044
                   Positive    Caryophyllene                    21.46         221.1899
                   Positive    Catechin                          3.44         291.0863
                   Positive    Chlorogenic acid                  3.11          355.084
                   Positive    Epigallocatechin              1.54 - 2.64      307.0812
                   Positive    Glutamic acid                     1.41         148.0604
                   Positive    Glutamine                         1.46          147.076
                   Positive    Kaempferol                       14.82         287.0552
                   Positive    Leucine                           1.76          132.101
                   Positive    Luteolin                         13.68         287.0551
                   Positive    Phenylalanine                     1.91          166.086
                   Positive    Proline                           1.49           116.07
                   Positive    Pyridoxine                        1.38         170.0812
                   Positive    Quercetin                        13.72         303.0498
                   Positive    Rhamnetin                        15.95         317.0653
                   Positive    Tryptophan                        2.49          205.097
                   Positive    Tyrosine                      1.54 - 1.77       182.081
                   Positive    Valine                            1.53          118.086
1418
1419
1420
1421
1422


                                                     43
1423   Table A.3. Results of the factorial ANOVAs of all stoichiometries and assigned metabolites
1424   extracted from Q. ilex leaves with vegetation structure (VS) and experimental drought
1425   treatment as factors for all seasons combined. The whole factorial model and the decomposed
1426   model are shown in the table for each of the known variables. Statistical significance (P < 0.05)
1427   is shown in red.
1428

                                                                    Decomposed model
                            Whole model
                                                              VS          Treatment    VStreatment
                             F       14.81       F           37.94            0.56             5.94
              C/N
                            (P)    (0.00000)    (P)        (0.00000)       (0.45658)        (0.01594)
                             F       15.36       F            6.78           28.28            11.03
              N/P
                            (P)    (0.00000)    (P)        (0.01012)       (0.00000)        (0.00112)
                             F       13.43       F            1.98           18.55            19.76
              C/P
                            (P)    (0.00000)    (P)        (0.16178)       (0.00003)        (0.00002)
                             F        4.64       F            4.52            3.27             6.14
              C/K
                            (P)    (0.00389)    (P)        (0.03510)       (0.07269)        (0.01429)
                             F        9.92       F            5.20            9.25            15.30
              K/P
                            (P)    (0.00001)    (P)        (0.02400)       (0.00276)        (0.00014)
                             F       10.30       F           15.94            3.80            11.15
              N/K
                            (P)    (0.00000)    (P)        (0.00010)       (0.05300)        (0.00105)
                             F       15.44       F           41.17            0.02             5.13
              [N]
                            (P)    (0.00000)    (P)        (0.00000)       (0.88056)        (0.02487)
                             F       15.42       F            3.37           24.19            18.69
               [P]
                            (P)    (0.00000)    (P)        (0.06831)       (0.00000)        (0.00003)
                             F        4.57       F            4.45            2.18             7.08
               [K]
                            (P)    (0.00426)    (P)        (0.03640)       (0.14197)        (0.00863)
                             F        5.67       F            4.91           10.74             1.35
               [C]
                            (P)    (0.00104)    (P)        (0.02821)       (0.00129)        (0.24648)
                             F        3.00       F            0.00            8.76             0.23
            Hexoses
                            (P)    (0.03254)    (P)        (0.96126)       (0.00357)        (0.63114)
                             F        8.26       F            6.03           10.03             8.73
           Pentoses
                            (P)    (0.00004)    (P)        (0.01515)       (0.00185)        (0.00362)
                             F        6.85       F           20.10            0.02             0.43
         Disaccharides
                            (P)    (0.00023)    (P)        (0.00001)       (0.88967)        (0.51054)
                             F        3.83       F            7.53            0.19             3.78
            Alanine
                            (P)    (0.01104)    (P)        (0.00677)       (0.66749)        (0.05362)
                             F        4.14       F            7.18            4.98             0.26
          Aspartic acid
                            (P)    (0.00745)    (P)        (0.00818)       (0.02713)        (0.61018)
                             F        1.77       F            3.81            0.51             1.00
         Glutamic acid
                            (P)    (0.15486)    (P)        (0.05271)       (0.47805)        (0.31933)
                             F        4.69       F            2.02            1.83            10.21
             Valine
                            (P)    (0.00366)    (P)        (0.15730)       (0.17776)        (0.00169)
                             F        2.88       F            1.30            5.82             1.51
            Leucine
                            (P)    (0.03801)    (P)        (0.25584)       (0.01702)        (0.22115)
                             F        3.22       F            6.02            0.68             2.96
         Phenylalanine
                            (P)    (0.02439)    (P)        (0.01528)       (0.40935)        (0.08737)
                             F        1.31       F            3.43            0.22             0.27
            Proline
                            (P)    (0.27473)    (P)        (0.06594)       (0.64043)        (0.60572)
                             F       13.45       F           20.43            0.41            19.50
            Arginine
                            (P)    (0.00000)    (P)        (0.00001)       (0.52359)        (0.00002)
          Tryptophan         F       0.46        F           0.12            0.85             0.40

                                                      44
                          (P)   (0.71267)   (P)        (0.72664)   (0.35927)   (0.52639)
                           F       2.66      F            6.50        0.36        1.10
          Tyrosine
                          (P)   (0.05040)   (P)        (0.01174)   (0.54711)   (0.29533)
                           F      11.64      F            0.14       34.76        0.00
           Adenine
                          (P)   (0.00000)   (P)        (0.70824)   (0.00000)   (0.96702)
                           F       2.51      F            7.24        0.28        0.00
         Adenosine
                          (P)   (0.06096)   (P)        (0.00791)   (0.59435)   (0.97502)
                           F       9.77      F           21.51        6.54        1.26
        Succinic acid
                          (P)   (0.00001)   (P)        (0.00001)   (0.01149)   (0.26316)
                           F       6.33      F           17.32        0.83        0.83
          Citric acid
                          (P)   (0.00045)   (P)        (0.00005)   (0.36286)   (0.36427)
                           F       7.07      F           12.41        2.30        6.50
          Malic acid
                          (P)   (0.00017)   (P)        (0.00056)   (0.13104)   (0.01175)
                           F       8.76      F           23.92        1.57        0.79
          Pyruvate
                          (P)   (0.00002)   (P)        (0.00000)   (0.21232)   (0.37513)
                           F       1.66      F            0.31        4.57        0.09
          Lactic acid
                          (P)   (0.17840)   (P)        (0.57576)   (0.03416)   (0.76162)
                           F      12.03      F            0.01       35.30        0.79
          Catechin
                          (P)   (0.00000)   (P)        (0.93463)   (0.00000)   (0.37528)
                           F       1.49      F            0.65        0.50        3.31
         Epicatechin
                          (P)   (0.21979)   (P)        (0.42025)   (0.47949)   (0.07076)
                           F      10.61      F            2.71        5.61       23.51
       Epigallocatechin
                          (P)   (0.00000)   (P)        (0.10183)   (0.01909)   (0.00000)
                           F       9.80      F           10.39        2.13       16.88
        Homoorientin
                          (P)   (0.00001)   (P)        (0.00154)   (0.14608)   (0.00006)
                           F       3.05      F            4.73        4.36        0.07
         Kaempferol
                          (P)   (0.03033)   (P)        (0.03122)   (0.03839)   (0.79761)
                           F       2.43      F            7.04        0.25        0.00
           Luteolin
                          (P)   (0.06734)   (P)        (0.00880)   (0.61678)   (0.98348)
                           F       5.67      F           10.21        3.91        2.89
          Quercitin
                          (P)   (0.00103)   (P)        (0.00169)   (0.04968)   (0.09140)
                           F       5.90      F           15.56        0.00        2.13
         Rhamnetin
                          (P)   (0.00077)   (P)        (0.00012)   (0.97022)   (0.14638)
                           F       8.85      F            5.64       10.18       10.74
       Chlorogenic acid
                          (P)   (0.00002)   (P)        (0.01877)   (0.00171)   (0.00129)
                           F       0.46      F            0.18        1.14        0.06
          Gallic acid
                          (P)   (0.71304)   (P)        (0.67443)   (0.28793)   (0.81406)
                           F      11.56      F            5.09       18.43       11.17
        -Humulene        (P)   (0.00000)   (P)        (0.02549)   (0.00003)   (0.00104)
                           F       8.27      F            8.89        6.25        9.65
        Caryophyllene
                          (P)   (0.00004)   (P)        (0.00332)   (0.01342)   (0.00224)
                           F       1.03      F            0.88        0.49        1.73
           Choline
                          (P)   (0.38100)   (P)        (0.34980)   (0.48699)   (0.19094)
                           F      29.10      F           59.86       17.87        9.56
         Pyridoxine
                          (P)   (0.00000)   (P)        (0.00000)   (0.00004)   (0.00235)
                           F       0.99      F            0.49        0.26        2.21
         Quinic acid
                          (P)   (0.40146)   (P)        (0.48713)   (0.60927)   (0.13934)
1429
1430




                                                  45
Table A.4. Results of the decomposed models of the factorial ANOVAs of all stoichiometries and assigned metabolites extracted from Q. ilex leaves with
vegetation structure (VS) and experimental drought treatment (TRT) as factors for each season. Statistical significance (P < 0.05) is shown in red.


                                  SUMMER                              AUTUMN                                   WINTER                              SPRING
                        TRT          VS       VSTRT        TRT          VS            VSTRT        TRT          VS       VSTRT        TRT          VS       VSTRT
                 F      0.02        5.29        1.18         1.67       21.29             4.42       0.08        12.56        0.54        0.06       22.21        7.92
    C/N
                (P)    (0.90)      (0.027)     (0.28)     (0.19968)   (0.00002)        (0.03882)   (0.7747)    (0.00111)   (0.46681)   (0.80691)   (0.00004)   (0.00786)
                 F      5.37         0.70        3.56        9.14        5.23             6.71       20.38        1.54        0.94        3.75        1.95        5.96
    N/P
                (P)   (0.0263)    (0.40846)   (0.06719)   (0.00342)   (0.02495)        (0.01151)   (0.00007)   (0.22312)   (0.33774)   (0.06079)   (0.17120)   (0.01972)
                 F       5.22        0.20        6.06        3.32        0.83            14.06       19.93        1.77        1.57        3.34        2.46       17.08
    C/P
                (P)   (0.02832)   (0.65922)   (0.01875)   (0.07236)   (0.36569)        (0.00034)   (0.00008)   (0.19220)   (0.21794)   (0.07582)   (0.12584)   (0.00020)
                 F       0.65        0.46        5.52        2.78        2.95             3.26        0.90        2.53        1.47        1.56        0.72        0.38
    C/K
                (P)   (0.42694)   (0.50328)   (0.02435)   (0.09955)   (0.09006)        (0.07514)   (0.34991)   (0.12025)   (0.23280)   (0.22035)   (0.40140)   (0.54178)
                 F       3.45        0.37       12.87        4.43        4.53             8.40        4.80        3.64        0.75        4.14        3.33        5.84
    K/P
                (P)   (0.07139)   (0.54952)   (0.00099)   (0.03852)   (0.03650)        (0.00491)   (0.03508)   (0.06436)   (0.39353)   (0.04935)   (0.07645)   (0.02082)
                 F       1.41        2.11        7.66        2.62       12.90             8.30        1.38        6.87        2.23        2.66        6.82        3.41
    N/K
                (P)   (0.24226)   (0.15485)   (0.00885)   (0.10945)   (0.00058)        (0.00514)   (0.24788)   (0.01279)   (0.14365)   (0.11182)   (0.01306)   (0.07291)
                 F       0.18        5.99        0.81        0.79       23.16             4.41       0.44        13.86        0.37        0.60       25.02        9.81
     [N]
                (P)   (0.67769)   (0.01935)   (0.37346)   (0.37622)   (0.00001)        (0.03912)   (0.5119)    (0.00067)   (0.54567)   (0.44463)   (0.00001)   (0.00344)
                 F       6.48        1.06        5.62       6.48         0.88            12.23       20.80        2.60        1.84        6.26        2.86       14.17
     [P]
                (P)   (0.01534)   (0.31024)   (0.02326)   (0.0129)    (0.35117)        (0.00079)   (0.00006)   (0.11552)   (0.18338)   (0.01704)   (0.09928)   (0.00060)
                 F       0.49        0.39        7.87        1.69        3.62             2.53        0.70        2.65        0.68        1.51        1.51        0.22
     [K]
                (P)   (0.48791)   (0.53847)   (0.00807)   (0.19754)   (0.06097)        (0.11613)   (0.40799)   (0.11229)   (0.41592)   (0.22754)   (0.22683)   (0.64182)
                 F       1.81        2.88        2.99        7.93        2.00             0.13        2.08        0.79        1.34        3.55        0.36        0.16
     [C]
                (P)   (0.18711)   (0.09852)   (0.09238)   (0.00619)   (0.16113)        (0.71668)   (0.15787)   (0.38043)   (0.25426)   (0.06752)   (0.55210)   (0.69474)
                 F       1.49        0.99        1.99        7.65        0.67             3.25        7.71        2.89        0.34        3.20        0.95        2.27
  Hexoses
                (P)   (0.22996)   (0.32744)   (0.16653)   (0.00714)   (0.41543)        (0.07549)   (0.00866)   (0.09769)   (0.56262)   (0.08210)   (0.33643)   (0.14057)
                 F       1.49        0.59        3.75        7.95        4.34             3.87        1.96        1.81        2.21       16.04        6.02        7.45
  Pentoses
                (P)   (0.22947)   (0.44667)   (0.06074)   (0.00613)   (0.04070)        (0.05290)   (0.17037)   (0.18696)   (0.14572)   (0.00030)   (0.01912)   (0.00976)
                 F       0.00        0.86        4.35        0.00       17.95             0.77        0.34       43.34        0.89        0.12       26.81        2.22
Disaccharides
                (P)   (0.96174)   (0.36043)   (0.04408)   (0.97119)   (0.00006)        (0.38258)   (0.56569)   (0.00000)   (0.35220)   (0.73313)   (0.00001)   (0.14482)
                 F       5.33        0.04        0.93        0.04        8.81             1.34       11.10        0.80        4.05        0.08        8.30        1.16
  Alanine
                (P)   (0.02687)   (0.83609)   (0.34045)   (0.85173)   (0.00401)        (0.25123)    (0.002)    (0.37662)   (0.05158)   (0.77374)   (0.00664)   (0.28883)
Aspartic acid   F       2.96        1.43        0.00        8.88       11.01             0.93        1.01        4.19        0.41        6.62        8.69        0.10




                                                                                  46
                   (P)   (0.09395)   (0.23958)   (0.98786)   (0.00387)   (0.00139)        (0.33794)   (0.32121)   (0.04809)   (0.52525)   (0.01435)   (0.00560)   (0.75819)
                    F       3.80        3.85        0.48        0.36       13.12             0.38        1.23        0.35        1.26        0.00       11.29        0.66
 Glutamic acid
                   (P)   (0.05923)   (0.05757)   (0.49307)   (0.54784)   (0.00052)        (0.54198)   (0.27546)   (0.55883)   (0.26969)   (0.99999)   (0.00186)   (0.42117)
                    F       3.19        0.00        3.52        0.03        1.86             1.28        7.03        1.05       22.92        0.12        1.83        0.57
    Valine
                   (P)   (0.08234)   (0.96725)   (0.06886)   (0.87114)   (0.17694)        (0.26058)   (0.01186)   (0.31208)   (0.00003)   (0.72827)   (0.18435)   (0.45378)
                    F      10.90        1.42        2.24        0.05        2.77             1.52        7.01        0.27        0.59        0.72        1.47        0.25
    Leucine
                   (P)   (0.00218)   (0.24189)   (0.14336)   (0.81675)   (0.10020)        (0.22127)   (0.01193)   (0.60653)   (0.44925)   (0.40108)   (0.23360)   (0.62315)
                    F       0.04        5.14        0.69        0.02        1.47            2.67         3.84        6.43        1.51        0.70        0.22        2.35
 Phenylalanine
                   (P)   (0.85115)   (0.02950)   (0.41204)   (0.88406)   (0.22951)        (0.1066)    (0.05795)   (0.01572)   (0.22641)   (0.40810)   (0.64427)   (0.13401)
                    F       1.77        0.97        0.12        0.52        0.87            0.32         0.04        6.45        6.57        0.48        0.10        1.15
    Proline
                   (P)   (0.19136)   (0.33127)   (0.72589)   (0.47378)   (0.35414)        (0.5743)    (0.83840)   (0.01553)   (0.01472)   (0.49218)   (0.75109)   (0.29162)
                    F       3.50        7.16        7.13        0.06        7.15            4.54         0.23        9.42       13.04        0.20        2.34        1.43
   Arginine
                   (P)   (0.06937)   (0.01117)   (0.01131)   (0.80970)   (0.00915)        (0.0364)    (0.63283)   (0.00407)   (0.00092)   (0.65685)   (0.13509)   (0.23999)
                    F       0.19       12.84        2.16        0.05        0.01             2.73        4.35        0.02        0.27        0.15        0.03        5.16
  Tryptophan
                   (P)   (0.66927)   (0.00100)   (0.15036)   (0.82535)   (0.92317)        (0.10244)   (0.04419)   (0.87845)   (0.60591)   (0.70002)   (0.85919)   (0.02920)
                    F      0.31        10.14        0.24        0.27        1.07             0.26        5.71        0.10        2.80        0.20        0.02        1.81
   Tyrosine
                   (P)   (0.5792)    (0.00299)   (0.62667)   (0.60408)   (0.30513)        (0.61135)   (0.02222)   (0.74873)   (0.10286)   (0.65685)   (0.89701)   (0.18666)
                    F      12.11        2.21        0.23        9.44        0.25             1.29       25.61        0.01        0.84        2.93        0.00        2.45
   Adenine
                   (P)   (0.00133)   (0.14616)   (0.63599)   (0.00295)   (0.61515)        (0.25893)   (0.00001)   (0.93598)   (0.36471)   (0.09548)   (0.95977)   (0.12620)
                    F       1.21        0.99        0.06        0.06        3.64             0.00        0.36        4.13        0.15        0.68        1.32        0.00
  Adenosine
                   (P)   (0.27862)   (0.32701)   (0.80169)   (0.81493)   (0.06028)        (0.98862)   (0.55060)   (0.04968)   (0.70189)   (0.41657)   (0.25839)   (0.96691)
                    F       2.38        4.79       4.56         4.55        8.37             0.37        0.12       14.06        2.38        8.64        3.47        0.68
 Succinic acid
                   (P)   (0.13169)   (0.03514)   (0.0396)    (0.03610)   (0.00497)        (0.54305)   (0.73180)   (0.00062)   (0.13139)   (0.00572)   (0.07068)   (0.41386)
                    F       0.79       37.26        0.67        0.09        5.70             1.64        6.07        1.68        0.68        1.01        7.40        1.99
   Citric acid
                   (P)   (0.37998)   (0.00000)   (0.41708)   (0.77076)   (0.01947)        (0.20441)   (0.01863)   (0.20303)   (0.41353)   (0.32186)   (0.01000)   (0.16697)
                    F       0.03       15.92        3.63        1.83        5.16             8.41        1.37        1.58        0.14        2.06        9.58        6.96
   Malic acid
                   (P)   (0.85751)   (0.00031)   (0.06484)   (0.18051)   (0.02596)        (0.00487)   (0.24945)   (0.21718)   (0.70657)   (0.15981)   (0.00380)   (0.01226)
                    F       0.47       13.26        0.00        0.27        7.02            0.94         1.88        9.56        0.06        2.07        4.50        0.46
   Pyruvate
                   (P)   (0.49817)   (0.00085)   (0.97975)   (0.60510)   (0.00980)        (0.3349)    (0.17902)   (0.00383)   (0.81333)   (0.15839)   (0.04095)   (0.50403)
                    F       1.39        0.11        0.22        0.58        0.13             0.22        4.23        0.07        0.39        1.37        0.00        0.07
  Lactic acid
                   (P)   (0.24690)   (0.73741)   (0.64378)   (0.44887)   (0.72193)        (0.63672)   (0.04703)   (0.79515)   (0.53534)   (0.24950)   (0.99999)   (0.78705)
                    F      18.55        1.53        0.00       11.15        0.03             0.43       10.86        1.45        0.74        4.06        1.22        0.00
   Catechin
                   (P)   (0.00012)   (0.22408)   (0.97594)   (0.00130)   (0.87445)        (0.51597)   (0.00222)   (0.23562)   (0.39602)   (0.05152)   (0.27626)   (0.97209)
                    F       0.96        0.28        5.19        0.75        0.75             0.57        3.94        0.01        0.07        0.17        2.18        0.19
  Epicatechin
                   (P)   (0.33376)   (0.59968)   (0.02870)   (0.39020)   (0.38936)        (0.45355)   (0.05470)   (0.90799)   (0.79156)   (0.67973)   (0.14821)   (0.66629)
                    F       0.96        0.43        6.01        5.87        0.87            14.02        0.02        1.78        3.13        4.46        1.06       13.15
Epigallocatechin
                   (P)   (0.33300)   (0.51388)   (0.01917)   (0.01783)   (0.35406)        (0.00035)   (0.88314)   (0.19046)   (0.08516)   (0.04167)   (0.31109)   (0.00088)




                                                                                     47
                    F       0.28        0.45        1.44        1.30        6.55             8.99        0.52        4.09        7.13        0.44        3.11        3.38
 Homoorientin
                   (P)   (0.59892)   (0.50837)   (0.23747)   (0.25754)   (0.01245)        (0.00366)   (0.47592)   (0.05062)   (0.01128)   (0.51034)   (0.08617)   (0.07408)
                    F       1.89        5.23        0.90        0.61        1.77             0.03        3.05        0.04        0.42        2.60        2.28        0.15
  Kaempferol
                   (P)   (0.17805)   (0.02817)   (0.34784)   (0.43879)   (0.18797)        (0.87326)   (0.08908)   (0.84802)   (0.52110)   (0.11581)   (0.13947)   (0.69812)
                    F       0.01        7.59        0.02        0.53        1.92             0.47        5.09        0.11        0.98        0.31        5.51        1.33
    Luteolin
                   (P)   (0.91111)   (0.00915)   (0.89501)   (0.46851)   (0.16968)        (0.49321)   (0.03031)   (0.74737)   (0.32992)   (0.58000)   (0.02455)   (0.25554)
                    F       0.31        5.31        0.95        2.21        2.81             1.60        1.58        2.58        0.36        2.80        1.30        0.77
   Quercitin
                   (P)   (0.58009)   (0.02711)   (0.33678)   (0.14159)   (0.09775)        (0.20917)   (0.21712)   (0.11689)   (0.55452)   (0.10322)   (0.26168)   (0.38461)
                    F       0.28        3.16        0.45        0.19        4.63             0.83        0.01        9.07        0.93        0.74        3.88        0.40
  Rhamnetin
                   (P)   (0.59808)   (0.08386)   (0.50820)   (0.66677)   (0.03461)        (0.36510)   (0.91982)   (0.00473)   (0.34148)   (0.39516)   (0.05647)   (0.53296)
                    F       6.74        7.18        1.64        0.34        0.69            18.53       16.22        0.05        2.77        3.24        0.78       18.13
Chlorogenic acid
                   (P)   (0.01356)   (0.01104)   (0.20889)   (0.56135)   (0.40882)        (0.00005)   (0.00028)   (0.82225)   (0.10468)   (0.08044)   (0.38193)   (0.00014)
                    F       0.54        0.94        0.10        0.82        0.09             0.45        0.00        0.62        0.00        0.39        3.45        0.17
   Gallic acid
                   (P)   (0.46657)   (0.33823)   (0.75092)   (0.36835)   (0.77082)        (0.50444)   (0.94698)   (0.43741)   (0.99585)   (0.53444)   (0.07164)   (0.68275)
                    F      3.05         0.35        0.16        8.00        3.75             9.25       15.56        3.42        6.28        6.21        4.10        7.54
 -Humulene
                   (P)   (0.0892)    (0.55577)   (0.68974)   (0.00597)   (0.05660)        (0.00323)   (0.00035)   (0.07280)   (0.01690)   (0.01742)   (0.05041)   (0.00936)
                    F       1.55        1.01       10.19        4.17        5.20            4.90         3.11        6.26        8.39        3.99        5.17        3.26
 Caryophyllene
                   (P)   (0.22177)   (0.32231)   (0.00293)   (0.04450)   (0.02546)        (0.0298)    (0.08641)   (0.01704)   (0.00637)   (0.05348)   (0.02911)   (0.07937)
                    F       4.99        1.08        0.10        7.83        0.00             2.38        0.01        0.61        0.05        2.85        0.45        4.54
    Choline
                   (P)   (0.03176)   (0.30599)   (0.74929)   (0.00650)   (0.97177)        (0.12669)   (0.93255)   (0.44102)   (0.82079)   (0.09982)   (0.50573)   (0.04001)
                    F       7.87       29.84        0.40       12.62       21.53            11.16        0.79       21.18        0.59        1.56       14.37        1.65
  Pyridoxine
                   (P)   (0.00806)   (0.00000)   (0.53193)   (0.00066)   (0.00001)        (0.0013)    (0.38009)   (0.00005)   (0.44933)   (0.21996)   (0.00055)   (0.20772)
                    F       0.31        0.85        2.47        2.35        0.33             0.00        4.06        0.05        4.98        7.73        1.22        0.04
  Quinic acid
                   (P)   (0.57817)   (0.36174)   (0.12462)   (0.12978)   (0.56506)        (0.97309)   (0.05142)   (0.82288)   (0.03195)   (0.00858)   (0.27700)   (0.83896)




                                                                                     48
1   Table A.5. P values of the one-way ANOVAs of coordinates of cases along PC1 and PC2 of the
2   experimental drought treatment and the vegetation structure of the PCAs for each season (Fig.
3   A.2).
4
      Season                                                                    PC axis       P
                                                                                 PC1         NS
                                       Treatment (Global)
                                                                                 PC2       0.0097
                                                                                 PC1         NS
                                       Structure (Global)
                                                                                 PC2       0.00001
                                                                                 PC1       0.0399
                                          For H-Forest (Control vs. Drought)
                     Treatment                                                   PC2       0.0027
      Winter
                  (within structure)                                             PC1       0.00001
                                           For L-Forest (Control vs. Drought)
                                                                                 PC2         NS
                                                                                 PC1       0.00000
                                          For Control (H-Forest vs. L-Forest)
                      Structure                                                  PC2       0.00000
                 (within treatment)                                              PC1        0.027
                                          For Drought (H-Forest vs. L-Forest)
                                                                                 PC2       0.00000
                                                                                 PC1       0.0065
                                       Treatment (Global)
                                                                                 PC2         NS
                                                                                 PC1         NS
                                       Structure (Global)
                                                                                 PC2       0.00000
                                                                                 PC1         NS
                                          For H-Forest (Control vs. Drought)
                     Treatment                                                   PC2         NS
       Spring
                  (within structure)                                             PC1       0.00037
                                           For L-Forest (Control vs. Drought)
                                                                                 PC2         NS
                                                                                 PC1       0.00000
                                          For Control (H-Forest vs. L-Forest)
                      Structure                                                  PC2       0.00000
                 (within treatment)                                              PC1       0.00000
                                          For Drought (H-Forest vs. L-Forest)
                                                                                 PC2       0.00000
                                                                                 PC1         NS
                                       Treatment (Global)
                                                                                 PC2         NS
                                                                                 PC1         NS
                                       Structure (Global)
                                                                                 PC2       0.00000
                                                                                 PC1       0.0267
                                          For H-Forest (Control vs. Drought)
                     Treatment                                                   PC2      0.073 (MS)
      Summer
                  (within structure)                                             PC1       0.0046
                                           For L-Forest (Control vs. Drought)
                                                                                 PC2         NS
                                                                                 PC1       0.00014
                                          For Control (H-Forest vs. L-Forest)
                      Structure                                                  PC2       0.00000
                 (within treatment)                                              PC1      0.098 (MS)
                                          For Drought (H-Forest vs. L-Forest)
                                                                                 PC2       0.00005
                                                                                 PC1         NS
                                       Treatment (Global)
                                                                                 PC2         NS
      Autumn                                                                     PC1      0.000051
                                       Structure (Global)
                                                                                 PC2       0.0033
                     Treatment            For H-Forest (Control vs. Drought)     PC1         NS


                                                            49
                       (within structure)                                             PC2   0.0756 (MS)
                                                                                      PC1     0.0029
                                                For L-Forest (Control vs. Drought)
                                                                                      PC2       NS
                                                                                      PC1    0.00000
                                                For Control (H-Forest vs. L-Forest)
                           Structure                                                  PC2   0.0886 (MS)
                      (within treatment)                                              PC1       NS
                                                For Drought (H-Forest vs. L-Forest)
                                                                                      PC2     0.0165
 5   NS, not significant; MS, marginally significant

 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33



                                                               50
34   Table A.6. R2 coefficients and the corresponding P values for the total, direct, and indirect
35   effects of the structural equation model (SEM) with tree diameter (D50), upper soil moisture
36   content, soil texture, and metabolomic variation (PC2) as endogenous variables. The model is
37   described in Fig. A.4.
38

        TOTAL EFFECTS
                               Drought                           Upper soil
                                             Soil texture                          DUSL            D50
                              treatment                          moisture
           Upper soil             0.2           -0.015                              0.056
           moisture           (P < 0.001)     (P = 0.42)                          (P = 0.26)

                                  0.39        -0.00074              0.048            0.64
              D50             (P < 0.0001)    (P = 0.45)          (P = 0.22)     (P < 0.0001)

              Soil                                                                 -0.097
            texture                                                               (P = 0.1)

                                  0.26         -0.0041                0.27           0.35           0.52
              PC2             (P < 0.0001)    (P = 0.42)         (P < 0.00001)   (P < 0.0001)   (P < 0.0001)


        DIRECT EFFECTS
                               Drought                           Upper soil
                                             Soil texture                          DUSL            D50
                              treatment                          moisture
           Upper soil             0.2           -0.015                              0.054
           moisture           (P < 0.001)     (P = 0.42)                          (P = 0.26)

                                  0.38                              0.048            0.64
              D50             (P < 0.0001)                        (P = 0.21)     (P < 0.0001)

               Soil                                                                -0.097
             texture                                                              (P = 0.1)

                                 0.004                               0.24                           0.52
              PC2              (P = 0.48)                         (P < 0.001)                   (P < 0.0001)


        INDIRECT EFFECTS
                               Drought                           Upper soil
                                             Soil texture                          DUSL            D50
                              treatment                          moisture
           Upper soil                                                              0.0015
           moisture                                                               (P = 0.44)

                                0.0098        -0.00074                             0.0027
              D50              (P = 0.24)     (P = 0.45)                          (P = 0.36)

              Soil
            texture
                                  0.26          -0.0041             0.025            0.35
              PC2             (P < 0.0001)     (P = 0.42)         (P = 0.22)     (P < 0.0001)
39
40
41

                                                            51
42   Fig A.1. Means ±SE for each vegetation structure for (a) trunk diameter (cm) measured at 50
43   cm from soil surface (D50) of the analyzed trees, (b) the depth of upper soil layers (DUSL) near
44   the analyzed trees and, (c) the aboveground biomass (AB)(Kg of dry weight) of Q. ilex trees in
45   the plots. A one way ANOVA was performed for each plot. F and P values are also shown.
46




47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65


                                                    52
66   Fig A.2. Principal component (PC) 1 vs. PC2 of the principal component analysis of the
67   metabolomic and stoichiometric variables for the Q. Ilex leaves for each season. Variable plots
68   for (a) winter, (b) spring, (c) summer, and (d) autumn, and case plots for (e) winter, (f) spring,
69   (g) summer, and (h) autumn. The variables in the variable plots are the same as in Figure 1 and
70   have the same codes. The case plots represent the cases by vegetation structure and
71   experimental drought treatment. Black arrows indicate the means for the PC1 and PC2 of the
72   corresponding vegetation structure (H-Forest or L-Forest) that differed significantly. Colored
73   arrows indicate the means for PC1 and PC2 of the corresponding treatment (control or
74   drought) in each vegetation structure that differed significantly. Different letters beside the
75   arrows indicate significant differences between seasons, vegetation structures, or
76   experimental treatments (P < 0.05).




                                                     53
          Formatat: Tipus de lletra: 8 pt
          Formatat: Centrada




77
     54
78   Fig A.3. Case plots of component 1 vs. component 2 of the partial least squares discriminant
79   analysis of the metabolomic and stoichiometric variables for the Q. ilex leaves for each season.
80




81
82
83
84
85
86
87
88
89




                                                    55
 90   Fig A.4. Structural equation model including the soil moisture, tree diameter (D50), soil
 91   texture, and metabolomic variation (PC2) as endogenous variables and depth of upper soil
 92   layers (DUSL), and experimental drought treatment as exogenous variables. The model
 93   coefficient (R2) and the corresponding P value for each relationship are shown between the
 94   two related variables. Total, direct, and indirect effects are shown in Table A.6.
 95
 96




 97
 98
 99
100

101




                                                      56