Quercus rotundifolia Bark as a Source of Polar Extracts: Structural and Chemical Characterization

: Quercus rotundifolia bark was studied regarding anatomical, chemical, and antioxidant properties from trees in two sites in southern Portugal and are here reported for the ﬁrst time. The general structure and anatomy of Q. rotundifolia bark showed a rhytidome with sequential undulated and anastomosed periderms with a small proportion of cork, while the phloem included broad rays with strong cell scleriﬁcation, groups of sclereids with embed large prismatic crystals, and abundant druses in parenchyma cells. The mean chemical composition was 15.5% ash, 1.6% dichloromethane extractives, 6.4% ethanol and 9.3% water extractives, 3.0% suberin, 30.5% total lignin, and 33.8% carbohydrates. Carbohydrates included mainly glucose (50.7% of total monomers) and xylose (23.8%), with uronic (3.0%) and acetic acids (1.0%). Suberin was mainly composed of ω -hydroxyacids (48.0% of all compounds) and α , ω -diacids (19.5%). The main compounds found in the lipophilic extracts were triterpenes (43.6%–56.2% of all compounds) and alkanoic acids (32.7%–41.7%). Phenolic content was high especially in the ethanol extracts, ranging from 219.5–572.9 mg GAE/g extract and comprising 162.5–247.5 CE/g extract of ﬂavonoids and 41.2–294.1 CE/g extract of condensed tannins. The extracts revealed very good antioxidant properties with IC 50 values of 4.4 µ g ethanol extract/mL and 4.7 µ g water extract/mL. Similar anatomical, chemical, and antioxidant characteristics were found in the bark from both sites. The high phenolic content and excellent antioxidant characteristics of polar extracts showed holm oak barks to be a promising natural source of antioxidants with possible use in industry and pharmaceutical/medical areas.


Introduction
The use of bark evolved from ancient times to present day, expanding according to the different socioeconomic contexts, as well as the scientific and technological advances. Barks show a large diversity and have a high chemical compound richness, namely regarding extractives such as sterols, terpenes, and a large number of different phenolic compounds, allowing application in medicine and pharmacy, adhesives, formaldehyde scavengers, and antioxidants.
Quercus rotundifolia Lam., generally known as holm oak ("azinheira" in Portuguese and "encina" in Spanish) due to its leaf's resemblance to Ilex aquifolium L. (the common European holly used in Christmas), is taxonomically complex and either recognized as a separate species or subspecies (Q. ilex subesp. ballota (Desf.) Samp. or Q. ilex subesp. rotundifolia (Lam.) O. Schwarz ex Table Morais) belonging to the subsection Sclerophyllodrys O. Schwartz [1,2]. It is naturally distributed in southern Europe (Portugal, south and southeast Spain) and northwestern Africa (mainly Morocco) in the western Mediterranean basin. Quercus rotundifolia is the main evergreen oak, besides Q. suber, which is characteristic of the Mediterranean typical agrosilvopastoral system in Portugal ("montado") and Spain ("dehesa"), that populates these savanna-like ecosystems. Quercus rotundifolia is found attention to the anatomical and chemical variability of the bark of different trees at two sites. These results will contribute to more knowledge-based decision-making on future Q. rotundifolia management within its natural and geographic distribution range as part of the montado.

Sites and Sampling
The bark samples were obtained from Q. rotundifolia trees selected along the species' natural distribution in Portugal-at the Perímetro Florestal da Contenda (38 • 03 N, 07 • 06 W; 450 m altitude; site 1), a stand under the management of the public institute ICNF (Instituto da Conservação da Natureza e das Florestas), and at Mora (38 • 56 N, 8 • 7 W; 130 m; site 2), a privately owned stand. For each site, legal permission was given to the sampling by ICNF. At both sites, the holm oak trees (hereafter referred to Q. rotundifolia, except if the opposite is mentioned) are sparse, the stands are unevenly aged and holm oaks are mixed with the dominant Q. suber trees. At each site, five trees were randomly selected and were sampled during February 2018 at site 1 and during October 2017 at site 2. The sampled trees are characterized in Table 1. The trees showed the characteristic holm oak architecture with an average of 7 m of tree height, a clear stem below branching of 1.6 m, and a 26.5 cm diameter at breast height (b.h., i.e., at 1.30 m above ground). Stand year plantation was not known and annual rings were not easily distinguishable in stem cross-sections (data not published). However, tree age was approximately estimated at 50-60 years and deemed to be similar at both sites. A 2 cm thick cross-sectional disc was cut from each tree at b.h. Bark thickness was measured along two cross-diameters. The samples of bark were manually removed, air-dried, and separated in two sets, one for anatomical characterization of the wood and the other for chemical analysis. Table 1. Characteristics of the studied Quercus rotundifolia trees from two sites (mean and standard deviation of five trees, min and max in parenthesis).

Cellular Structure Characterization
The macroscopic observations were made on the bark sample cross-section after surface sanding (P 1000) using a modular stereomicroscope (Leica MZ6, Leica Microsystems Ltd., Heerburg, Germany) coupled to a digital camera (Leica DC320, Leica Microsystems Ltd., Heerburg, Germany). For the microscopic observations, the bark samples were impregnated with polyethylene glycol (DP 1500), and transverse and longitudinal microscopic sections of approximately 17 µm thickness were prepared with a sliding microtome (Leica SM 2400, Leica Microsystems Nussloch GmbH, Nussloch, Germany) using adhesive for sample retrieval. The sections were stained with a double staining of chrysodine/astra blue and Sudan 4 was used for selective staining of suberin. Individual bark specimens were also macerated for observation. Slide preparation and maceration followed standard procedures described in previous works [28]. Phloem and rhytidome were measured at two random intact points using image analysis systems (Leica Qwin Pro, v 3.5.0). Qualitative and quantitative observations were made using light microscopy (Nikon Microphot-FXA, Nikon, Japan). Bark terminology followed the IAWA List of Microscopic Bark Features [33].

Chemical Summative Analysis
The bark samples of each tree were ground separately in a cutting mill (Retsch SM 2000, Retsch GmbH, Haan, Germany) using an output sieve of 10 mm × 10 mm, followed by one of 2 mm × 2 mm, and fractionated with a vibratory system with standard sieves (Retsch AS 200, Retsch GmbH, Haan, Germany). The 40-60 mesh (0.425-0.250 mm) fractions were used for chemical analysis. The summative chemical analysis included determination of ash; extractives in dichloromethane, ethanol, and water; suberin; Klason and acid-soluble lignin; and the monomeric composition of polysaccharides. Determinations were made in duplicate samples. The methods followed the procedures adopted in our laboratory for bark chemical characterization (e.g., [23,34]) and can be briefly described as follows. The ash content was determined by incinerating 2.0 g of each sample at 525 • C overnight and weighing the residue, reported as percentage of the original samples (Tappi 211 om-02). The extractives were determined with procedures adapted from Tappi 204 cm-97, performed in a Soxhlet system with dichloromethane, ethanol, and water under reflux successively, after which the content was calculated for each solvent by mass difference of the ovendried solid residue and reported as a percentage of the original sample. The suberin content was determined by methanolysis for depolymerization using 1.5 g of the sample of extractive-free material and is expressed as a percentage of the initial dry mass [35]. The lignin content was analyzed from the extracted and desuberinized samples by acid hydrolysis. Klason lignin was determined as the mass of the solid residue after drying at 105 • C (Tappi T 222 om-02). The acid-soluble lignin was determined by measuring the UV absorbance (TAPPI Useful Method UM 250). The remaining acid solution was kept for sugar analysis. The carbohydrates were calculated based on the amount of the neutral sugar monomers (rhamnose, arabinose, xylose, galactose, mannose, and glucose) and uronic acids (galacturonic and glucuronic acids) released by total hydrolysis, after derivatization as alditol acetates and separation by high-pressure ion-exchange chromatography with a pulsed amperometric detector (HPIC-PAD). The content of acetic acid was also determined in the hydrolysate using high-pressure ion-exclusion chromatography with a UV/visible detector (HIPCE-UV).

Composition and Antioxidant Activity of Polar Extracts
The ethanol and water extracts were obtained by successive Soxhlet extraction and analyzed in relation to total phenolics (TPC), flavonoids (FC), and condensed tannin (TC) content, as previously described [23]. Each assay was performed at least three times and at least three independent replicates were prepared for each standard and sample. The Folin-Ciocalteu method was used for TPC determination, using gallic acid as standard, and the results were reported as mg gallic acid equivalents (GAE)/g of dried bark extract. The AlCl 3 colorimetric assay was used for the FC determination, using catechin as standard, with the results expressed as mg of catechin equivalents (CE)/g of dried bark extract. TC was determined by the vanillin-sulfuric acid assay, using catechin as standard, and the results are expressed as mg catechin equivalents (CE)/g of dried bark extract. The antioxidant activity of the ethanol and water extracts was determined using two methods-ferric reducing/antioxidant power (FRAP), which measures the sample's ferric reducing power, and 2,2-diphenyl-1-picryhydrazyl (DPPH), which measures the free radical scavenging capacity, as previously described [23]. FRAP is expressed as Mmol Trolox equivalents/g dry mass and the DPPH is expressed in terms of the amount of extract required to reduce 50% of the DPPH concentration (IC 50 ) and Trolox equivalents on a dry extract base (TEAC).

Composition of Lipophilic Extracts
Aliquots of the dichloromethane (DCM) extracts (5 mL) were taken, evaporated under N 2 flow, and dried at room temperature under vacuum overnight. For the GC-MS analysis, one aliquot from the DCM extracts (2 mg) was derivatized to trimethylsilyl ethers/esters (TMS) with 100 µL of pyridine (Sigma-Aldrich, St. Louis, MO, USA) and 100 µL of BSTFA (N,O-bis(trimethylsilyl)trifluoroacetamide with 1% trimethylchlorosilane, Sigma-Aldrich, St. Louis, MO, USA) at 60 • C for 30 min. After cooling at room temperature, the extracts were injected in splitless mode in a GC-MS (Agilent 7890 A-5975C MSD, Santa Clara, CA, USA) with an autoinjector and a high-temperature capillary column (Zebron 5 H T, 30 m × 0.25 mm x 0.1 µm, Anaheim, CA, USA) using He as carrier gas at a constant flow of 1 mL/min. The injector temperature was 280 • C and the oven was programmed with an initial temperature of 100 • C (1 min), 10 • C/min to 250 • C (1 min), 8 • C/min to 350 • C (5 min), and 8 • C/min to 380 • C (5 min). The MS source conditions were MSD transfer-line temperature maintained at 330 • C, the MS source at 230 • C, the quadrupole at 150 • C, and the electron ionization energy at 70 eV. The electronic impact mass spectra (EIMS) were taken over a range of m/z 40-950. For semi-quantification analysis, the GC-MS was externally calibrated with standards (hexadecanoic acid and asiatic acid) that are representative of the major families of the lipophilic extracts (saturated fatty acids and triterpenes, respectively). The respective multiplication factors needed to acquire a correct quantification of the peak areas were calculated as an average of three GC-MS runs. The compounds are expressed as a % of each peak in TIC. Each aliquot was injected and duplicated. The identification of the compounds (as TMS derivatives) was based on comparisons with standards, Wiley 6 and NIST libraries data, and interpretation of mass spectrometric fragmentation patterns.

Composition of Suberin
Aliquots of the dichloromethane extracts (5 mL) from the suberin depolymerization reaction were taken, evaporated under N 2 flow, and dried at room temperature under vacuum overnight. For the GC-MS analysis, one aliquot from the DCM extracts (1 mg) was derivatized to trimethylsilyl ethers/esters (TMS) with 100 µL of pyridine (Sigma-Aldrich, St. Louis, MO, USA) and 100 µL of BSTFA (N,O-bis(trimethylsilyl) trifluoroacetamide with 1% trimethylchlorosilane, Sigma-Aldrich, St. Louis, MO, USA) at 60 • C for 30 min. The subsequent procedures are described above.

Statistical Analysis
All results are expressed as mean and standard deviation. The significance of differences (p < 0.05) among the corresponding mean values was determined by one-way analysis of variance (ANOVA) using the SPSS statistical software (version 26).

Structure and Anatomy
The external appearance of the bark of Q. rotundifolia was finely square fissured. It was similar in the trees at both sites, with 11.6 mm mean width, including equally distributed rhytidome and secondary phloem (Table 1 and Figure 1). The rhytidome of Q. rotundifolia was persistent, with 5.1 mm mean width (Table 1 and Figure 1), and composed of various periderms and dead secondary phloem between them (Figure 2A-C). Its qualitative anatomy did not differ among trees. The sequential periderms formed thin and undulated layers that occasionally anastomosed ( Figure 2A); in each periderm, the phellem (cork) was poorly developed, comprising layers (<10 layers) The rhytidome of Q. rotundifolia was persistent, with 5.1 mm mean width (Table 1 and Figure 1), and composed of various periderms and dead secondary phloem between them (Figure 2A-C). Its qualitative anatomy did not differ among trees. The sequential periderms formed thin and undulated layers that occasionally anastomosed ( Figure 2A); in each periderm, the phellem (cork) was poorly developed, comprising layers (<10 layers) of radially flattened cells arranged into a more or less distinct radial pattern with thin suberized walls or, sometimes, lignified thick walls. The phelloderm consisted of a few layers (1-3) of rectangular to round cells, also with radial alignment ( Figure 2D). Both phellem and phelloderm cells may be filled with dark contents, presumably phenolic compounds. The phellogen, which gives rise to the phellem outside and phelloderm inside, is, in general, difficult to recognize but sometimes could be observed as rectangular and thin-walled cells in cross section ( Figure 2D). The rhytidome of Q. rotundifolia was persistent, with 5.1 mm mean width (Table 1 and Figure 1), and composed of various periderms and dead secondary phloem between them (Figure 2A-C). Its qualitative anatomy did not differ among trees. The sequential periderms formed thin and undulated layers that occasionally anastomosed ( Figure 2A); in each periderm, the phellem (cork) was poorly developed, comprising layers (<10 layers) of radially flattened cells arranged into a more or less distinct radial pattern with thin suberized walls or, sometimes, lignified thick walls. The phelloderm consisted of a few layers (1-3) of rectangular to round cells, also with radial alignment ( Figure 2D). Both phellem and phelloderm cells may be filled with dark contents, presumably phenolic compounds. The phellogen, which gives rise to the phellem outside and phelloderm inside, is, in general, difficult to recognize but sometimes could be observed as rectangular and thin-walled cells in cross section ( Figure 2D). The secondary phloem (Figures 3-5) included the conducting and nonconducting phloem, with a gradual transition between these two layers. The phloem was nonlayered and the growth rings were difficult to recognize. The secondary phloem (Figures 3-5) included the conducting and nonconducting phloem, with a gradual transition between these two layers. The phloem was nonlayered and the growth rings were difficult to recognize.
The conducting phloem near the vascular cambium represented a narrow portion of the entire phloem ( Figure 3A) and consisted of the functional sieve tube elements and companion cells (conducting tissue), the axial parenchyma (storied tissue), rays (storage/transversal conduction), and fibers (mechanical support). The sieve tubes elements were thin, nonlignified walled cells with round to irregular shape in transverse view and were mostly tangentially or radially grouped ( Figure 3D). The sieve tube elements had compound and inclined sieve plates and numerous lateral sieve areas ( Figure 3E). The companion cells were recognized in transversal sections ( Figure 3D). The axial parenchyma cells were nonlignified and thin-walled with a round shape of irregular size in transverse view and appeared as thin strands of a few cells located between fibers and interspersed with sieve elements (Figures 3D and 4B). The fibers were arranged in small groups as discontinuous tangential bands crossed by thin rays ( Figure 2C); fibers were thick-walled and lignified, accompanied by chambered crystalliferous cells. (Figures 4A,B and 5A). The rays were comprised of two types (Figures 3A-C, 4B and 5A,B): uniseriate rays and broad rays (multiseriate, up to 20 cells wide and >100 cells high), both homocel-lular with procumbent cells ( Figure 5C). These broad rays often protrude into the xylem ( Figure 3B). Sclerification of radial parenchyma cells occurred mostly in these fused broad rays ( Figures 3A,B and 5B), early in the conducting phloem near the cambium ( Figure 3B).       The conducting phloem near the vascular cambium represented a narrow portion of the entire phloem ( Figure 3A) and consisted of the functional sieve tube elements and companion cells (conducting tissue), the axial parenchyma (storied tissue), rays (storage/transversal conduction), and fibers (mechanical support). The sieve tubes elements were thin, nonlignified walled cells with round to irregular shape in transverse view and were mostly tangentially or radially grouped ( Figure 3D). The sieve tube elements had compound and inclined sieve plates and numerous lateral sieve areas ( Figure 3E). The companion cells were recognized in transversal sections ( Figure 3D). The axial parenchyma cells were nonlignified and thin-walled with a round shape of irregular size in transverse view and appeared as thin strands of a few cells located between fibers and interspersed with sieve elements (Figures 3D and 4B). The fibers were arranged in small groups as discontinuous tangential bands crossed by thin rays ( Figure 2C); fibers were thick-walled and lignified, accompanied by chambered crystalliferous cells. (Figures 4A,B and 5A). The rays were comprised of two types (Figures 3A-C, 4B and 5A,B): uniseriate rays and broad rays (multiseriate, up to 20 cells wide and >100 cells high), both homocellular with procumbent cells ( Figure 5C). These broad rays often protrude into the xylem ( Figure 3B). Sclerification of radial parenchyma cells occurred mostly in these fused broad rays ( Figures 3A,B and 5B), early in the conducting phloem near the cambium ( Figure 3B).
The nonconducting tissue started with the collapse of the sieve elements. At this point, the structure of the secondary phloem became disorganized with distortion and expansion of cells with subsequent sclerification, associated with the dilatation growth. Sclereids and fiber-sclereids ( Figure 6A) with thick and heavily lignified cells walls appeared and often formed spherical or irregular prominent groups ( Figures 3A and 6A-C), dispersed in the phloem adjacent to the fibers and near or within the broad rays. Abundant sclerification of parenchyma cells occurred in the outer portion of phloem near the inner periderm. The nonconducting tissue started with the collapse of the sieve elements. At this point, the structure of the secondary phloem became disorganized with distortion and expansion of cells with subsequent sclerification, associated with the dilatation growth. Sclereids and fiber-sclereids ( Figure 6A) with thick and heavily lignified cells walls appeared and often formed spherical or irregular prominent groups ( Figures 3A and 6A-C), dispersed in the phloem adjacent to the fibers and near or within the broad rays. Abundant sclerification of parenchyma cells occurred in the outer portion of phloem near the inner periderm.  Crystals, presumably of calcium oxalate, were abundant through all the bark, mostly present as abundant druses (Figures 3D and 6B,C) in axial parenchyma adjacent to sieve tube elements and as prismatic crystals in both sclereids (one large crystal/sclereid, Figures 2C, 4C, and 6B) and axial parenchyma cells, e.g., crystalliferous parenchyma ( Figure  6B), which bordered the fiber groups ( Figures 4B and 5A). Abundant phenolic compounds were observed by dark color staining in the secondary phloem and rhytidome e.g., axial and ray parenchyma cells, in the sclereids, phellem, and phelloderm cells (Figures 2B-D and 4A-C). Crystals, presumably of calcium oxalate, were abundant through all the bark, mostly present as abundant druses (Figures 3D and 6B,C) in axial parenchyma adjacent to sieve tube elements and as prismatic crystals in both sclereids (one large crystal/sclereid, Figures 2C, 4C and 6B) and axial parenchyma cells, e.g., crystalliferous parenchyma ( Figure 6B), which bordered the fiber groups ( Figures 4B and 5A). Abundant phenolic compounds were observed by dark color staining in the secondary phloem and rhytidome e.g., axial and ray parenchyma cells, in the sclereids, phellem, and phelloderm cells ( Figures 2B-D and 4A-C).

Chemical Composition
The summative chemical compositions of the Q. rotundifolia bark samples, from the two sites, are summarized in Table 2. The mean composition was (in % of the oven dry bark) 15.5% ash, 17.2% extractives, 3.0% suberin, 30.5% lignin, and 33.8% polysaccharides. Q. rotundifolia bark showed a high content of extractives of 15.1% (site 2) and 19.3% (site 1). The main contribution came from polar compounds solubilized by ethanol and water, representing 88% and 93% of the total extractives, respectively, for barks from trees of site 2 and 1; the non-polar compounds extracted by dichloromethane corresponded only to an average of 10.2% of the total extractives. Suberin content was low at 2.9% (site 2) and 3.1% (site 1), while the total lignin content was relatively high at 29.1% (site 1) and 31.8% (site 2). The ash content of Q. rotundifolia bark samples was particularly high. Between sites, a statistically significant difference was detected only in the content of dichloromethane and ethanol extractives. The carbohydrate composition is summarized in Table 3 in regards to the proportion of neutral monosaccharides, acetates, and uronic acids. The major monosaccharide was glucose, corresponding to 52% and 49% of the total monomers, while xylose was the dominant non-cellulosic sugar with 24%; arabinose, galactose, and rhamnose were also present (9.3%, 5.4%, and 3.3%, respectively). Uronic acids were also present, representing 5.9% of the total content of monomers as well as acetyl groups (1.0%). The carbohydrate composition was similar at both sites.

Polar Extracts Composition
The results for TPC, FC, and TC, as well as the antioxidant activity of the extracts obtained successively with ethanol and water, are shown in Table 4. The ethanol extracts contained much higher values of phenolics, flavonoids, and tannins than the subsequent water extracts. The bark ethanol extracts had a high proportion of phenolic compounds (572.8 mg GAE/g extract; 3.7 g GAE/g dry bark), in which flavonoids and condensed tannins constituted the major classes (247.6 mg CE/g extract and 294.1 mg CE/g extract, respectively). The bark water extracts contained a much lower amount of phenolic compounds: total phenolics 219.5 mg GAE/g of extract, flavonoids 162.5 mg CE/g of extract, and condensed tannins 41.2 mg CE/g of extract. Significant differences were found between sites for the ethanol and water extracts. The radical scavenging activities of the bark extracts corresponded to IC 50 values of 4.4 µg extract/mL and 4.7 µg extract/mL, respectively, for the ethanol and water extracts. The reducing ability of the ethanol extracts by the FRAP assay was, on average, 4.29 mM Trolox/g of extract, while that of the water extracts was on average 1.34 mM Trolox/g of extract. The antioxidant capacity and the reducing ability of the polar extracts were significantly different between sites.

Structure and Anatomy
The finely square fissured external appearance of the bark of Q. rotundifolia (Figure 1) is in accordance with the available description [36]. The bark is quite different from that of other oaks, and its grey to dark color is also a distinctive characteristic. Bark thickness is highly variable within the species, namely in relation to tree age; a similar bark thickness (2-6 mm) was reported at 1.5-1.8 m of stem height for 50-100 years old trees [32]. The holm oak bark thickness and the rhytidome were narrower compared to Q. petraea (10.5-29.1 mm and 3.7-23.3 mm, respectively) [37] and similar to Q. faginea (5 mm to 14 mm) [28]. The anatomical features of the secondary phloem of the bark of Q. rotundifolia ( Figures 3A-E, 4A-C, 5A-C and 6A-C) were similar to those of other Quercus species: Q. robur [38,39], Q. sessiliflora [40], Q. petraea [37,41], Q. faginea [28], Q. cerris [29], Q. infectoria, Q. alnifolia, and Q. rubra [36], and other ten Quercus species [42]. The main similarities were in the type of rays (uni/multiseriate), the strong sclerification of cells in the broad multiseriate rays, the occurrence and formation of large groups or clusters of sclereids, and the location of crystals (prismatic crystals and druses). The pattern of rhytidome development with thin sequential periderms in Q. rotundifolia (Figure 2A) was similar to that in various other Quercus species, leading to a low proportion of phellem in these barks (Figure 2A,B,D) and corroborating early findings from holm oak growing in Portugal [30]. Among oaks, one exception is the cork-rich bark of Q. cerris, which may have a substantial proportion of phellem in the rhytidome [29], and Q. suber [13] and Q. variabilis [15,43], with only one periderm with a continuous and thick cork layer.

Chemical Composition
The high extractives content and their substantial polar fraction were in line with the anatomical observations of phenolic deposits in the secondary phloem and rhytidome cells ( Figures 1B-D and 3A-C). This result aligns with results found in barks of other Quercus spp., such as Q. faginea (total extractives 13.2%, with an 85.6% proportion of ethanol and water extractives) [23], Q. rubra (12.1%, with 9.4% of ethanol-water extract) and Q. robur (23.0%, with 21.9% of ethanol-water extract) [26], Q. laurina (14.2% hot water and 13.6% ethanol extractives), Q. crassifolia (20.7% hot water and 11.0% ethanol extractives), and Q. scytophylla (6.8% hot water and 4.4% ethanol extractives) [44], or Q. laurina (19.3% total extractives) and Q. crassifolia (12.7% total extractives) [25]. The polar extractives content was particularly high at site 2 and the significant differences between sites might be related with the effect of growth season sampling time and/or drought on the formation of these secondary metabolites. In a recent study, Leite et al. [16] analyzed the effect of drought on cork chemical composition and showed that drought enhanced the amount of the polar extractives soluble in ethanol.
The ash content was very high (Table 2), which is in accordance with the abundant crystals observed mostly in sclereids and in axial parenchyma adjacent to fiber and sieve tube elements.

Polar Extract Composition
The bark of Q. rotundifolia was very rich in phenolic compounds that may be solubilized by polar solvents. When considering the extraction of polar compounds successively with ethanol and water, the results showed that most of the phenolics were extracted by ethanol (Table 4), mostly constituted by flavonoids (247.4 mg CE/g extract) and condensed tannins (294.1 mg Ceq/g extract). The subsequent water extraction solubilized the remaining phenolic compounds with much lower proportions of flavonoids (162.5 mg Ceq/g extract) and condensed tannins (41.2 mg Ceq/g extract). Similar reports on high TPC have been given in the literature for Quercus spp. bark polar extracts, even if the extraction procedures differed. For ethanol:water extracts (50:50), Ferreira et al. [23] reported for Q. faginea bark 630.3 mg GAE/g of extract of total phenols, 207.7 mg CE/g of extract of flavonoids and 220.7 mg CE/g of extract of condensed tannins, and Sillero et al. [26] for Q. rubra and Q. robur 276.5-610.6 mg GAE/g of extract of total phenolics and 650.4-1021.8 mg CE/g of extract of flavonoids. Valencia-Avilés et al. [44] compared the ethanol and hot water extraction for Q. laurina, Q. crassifolia, and Q. scytophylla barks and reported high TPC (from 329 to 756 mg GAE/g extract) but lower flavonoids (12.9 to 25.4 mg QE, (quercetin)/g extract) and condensed tannins (12.6 to 53.5 mg CChE (cyanidin chloride equivalents)/g extract).
The Q. rotundifolia bark extracts showed very good antioxidant properties ( Table 4). The IC 50 values demonstrated this high antioxidant activity (IC 50 = 4.4 µg ethanol extract/mL and 4.7 µg water extract/mL) when compared to the antioxidant standard Trolox (IC 50 = 3.3 and 3.6 µg Trolox/mL, in ethanol and water, respectively). Sillero et al. [26] and Ferreira et al. [23] reported very similar antioxidant properties of ethanol:water bark extracts of Q. rubra and Q. robur (399.62 mg TEAC/g extract and 1521.25 mg TEAC/g ex-tract, respectively) and Q. faginea (1576.12 mg TEAC/g extract; IC 50 = 2.63 µg extract/mL). Santos et al. [47] reported for Q. suber cork an IC 50 value of 2.8 µg water extract/mL, 3.6 µg methanol extract/mL and 5.8 µg methanol-water extract/mL. The results of the FRAP assay follow a similar trend to the one observed in the results obtained for DPPH. The ethanol extracts showed higher FRAP activity than the water extracts with 4.3 mM Trolox/g of extract and 1.3 mM Trolox/g of extract, respectively. In a sequential extraction, water extracts had lower antioxidant properties than ethanol extracts, but they still had compounds with antioxidant capability that were not solubilized in ethanol. The reducing ability by FRAP of both extracts was similar to the values of 4.44 mM TEAC/g of extract for Q. faginea bark ethanol:water (50:50) extracts [23].

Lipophilic Extracts Composition
The compositions of lipophilic extractives of Q. rotundofolia bark are shown in Table 5. The main constituents were triterpenes (49.9% of all compounds), namely ursolic, betulinic, and oleanolic acids, as well as betulin, with quantitative differences between the two sites. At site 1, the triterpenic acids were much more relevant than betulin, with ursolic, betulinic, and oleanolic acids constituting 35.2%, 33.4%, and 13.3% of all triterpenes, respectively, whereas betulin was only found in small amounts (3.1% of all triterpenes). However, the opposite was observed in the Q. rotundifolia bark samples collected at site 2-oleanolic acid was found in smaller amounts (only 9.3% of all triterpenes) and ursolic acid comprised 18.8% of all triterpenes identified, whereas betulinic acid and betulin represented 21.9% and 23.7% of all triterpenes, respectively. Other Quercus species showed different lipophilic compositions, e.g., Q. suber cork also included triterpenes belonging to friedelane, lupine, and steroid families, as well as long alkanoic chains [48], but in different amounts; Q. cerris cork also showed high amounts of botulin [49]; and Q. variabilis cork also showed a high triterpenic content, representing around 52.0% of all compounds [43]. Friedelin was found in a fair quantity in the dichloromethane extracts of Q. rotundofolia bark (between 1.8% and 5.0%), which was slightly higher than the reported for Q. suber cork by Castola et al. ([48,50]) and Sousa et al. [51], while it was the main constituent of Q. cerris cork (representing 6.2%) [14]. Q. rotundofolia bark lipophilic extracts also included long-chain lipid compounds e.g., saturated alkanoic acids (23.7%-31.2% of all compounds), constituting 72.4%-74.0% of all long-chain acids. Substituted alkanoic acids, α,ω-diacids and ω-hydroxyacids also existed but in smaller amounts. Similar results have been reported in the literature for Q. faginea barks, where saturated alkanoic acids also constituted an abundant group of compounds (around 25.0%), with hexadecanoic acid and docosanoic acid as the most representative elements of this family of compounds [23].

Suberin Composition
The results for the suberin composition obtained by GC-MS analysis are summarized in Table 6. The main suberin monomers of Q. rotundifolia bark were fatty acids, representing 84.0% of all compounds, in the samples collected at both sites. Among them, ω-hydroxyacids were the major compounds, representing 46.8% and 50.2% in the samples from site 1 and 2, respectively. The most abundant compounds were 9,10,18trihydroxyoctadecanoic acid at site 2 and 18-hydroxy-9-octadecenoic acid at site 1, representing 31.3% and 29.3% of all ω-hydroxyacids, respectively. α,ω-Diacids (saturated and substituted) were also abundant, with hexadecanedioic acid and 9,10-dihydroxyoctanedioic acid as the most representative at rates of 25.7% and 19.4% of all α,ω-diacids at site 1 and 28.1% and 38.7% α,ω-diacids at site 2, correspondingly. The literature on suberin composition of other oaks shows that it is species specific, i.e., it varies between Quercus species. For instance, in Q. faginea bark, suberin is mainly composed by fatty acids, namely ω-hydroxyacids (between 40%-50% of all compounds) and substituted α,ω-alkanoic diacids [23]. In Q. suber, for instance, the main constituents of suberin are substituted α,ω-diacids with mid-chain epoxy or diol substitutions [52]. In Q. cerris, however, ωhydroxyacids represent 90% of the long chain monomers [45], as well as in Q. variabilis where ω-hydroxyacids are also the major compounds identified (around 58.7%), followed by α,ω-substituted diacids (19.5%) [43]. Alkanoic acids, aromatics, and alkanols are also present, in amounts varying from around 4.0%-19.0%. Trace amounts of dehydroabietic acid and β-sitosterol were also identified, which was somewhat common in Q. faginea barks, where no sterols or triterpenes were identified, as reported by Ferreira et al. [23].

Conclusions
The general structure and anatomy of Q. rotundifolia bark were described for the first time and proved similar to most Quercus spp. The main pattern of Q. rotundifolia bark showed a rhytidome with sequential undulated and anastomosed periderms with a small proportion of cork, while the phloem included broad rays with strong cell sclerification, groups of sclereids with embedded large prismatic crystals, and abundant druses in parenchyma cells. The chemical composition of Q. rotundifolia bark was also studied for the first time, including variability of composition of lipophilic and polar extracts, and antioxidant activity, as well as suberin composition. The polar extractive content was high, especially for ethanol extracts in regards to total phenolics, flavonoids, and condensed tannins. The Q. rotundifolia bark extracts showed very good antioxidant properties and therefore should be considered as a relevant natural source. Certainly, further studies are needed for the identification of single compounds and to gather information on their structural elucidation and their biological activity. Suberin levels were low, in accordance with the bark anatomical characteristics, mainly composed by fatty acids and substituted α,ω-alkanoic diacids. as in various other Quercus spp., but also confirming the betweenspecies variation. These findings constitute the grounds for the valorization of Q. rotundifolia bark, which may be integrated into a biorefinery approach with a first step of polar extracts fractionation, thereby contributing to local communities' economies and to the overall sustainability of this species and the montado ecosystem in the western Mediterranean area.