• Among oak species, Quercus ilex is classified as a monoterpene emitter and Q. suber is mainly known as a nonisoprenoid emitter. The extent and origin of this diversification is unknown.
• We examined intra- and interspecific emission variability in two mixed stands which differed in their level of hybridization and reciprocal genetic introgression based on variations in cytoplasmic (chloroplast DNA) and nuclear (allozyme) markers.
• At both sites all trees identified as Q. ilex, or as recent descendants from Q. ilex × Q. suber hybrids, emitted monoterpenes. Of Q. suber trees (genetically introgressed or not by Q. ilex), 91% were also monoterpene emitters, and the remainder nonemitters. One tree identified as a Q. canariensis × Q. ilex hybrid emitted both isoprene and monoterpenes. Compared with Q. ilex, the standard emission rate of Q. suber was higher in summer and lower in autumn. Both species emitted the same monoterpenes, proportions of which showed significant intra- and interspecific variability.
• The results suggest that Q. suber populations in the French Mediterranean intrinsically emit monoterpenes, and that gene flow between oak species contributes to diversification of emission signatures.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Emissions of volatile organic compounds (VOC) from terrestrial vegetation are of considerable interest to atmospheric scientists because they interact in air pollution events and affect climate by influencing the atmosphere's oxidative capacity and by contributing to particle formation and growth (Kesselmeier & Staudt, 1999; Monson & Holland, 2001; Sharkey & Yeh, 2001). Air chemistry and climate models need large-scale emission inventories that estimate the annual load of phytogenic VOCs into the atmosphere on regional and global scales. These estimations are highly uncertain, largely because of our limited knowledge of the genetic diversity of emission sources.
Oaks have been recognized as among the greatest VOC emitters (Harley et al., 1999). Inside the chloroplasts their leaves produce large amounts of volatile isoprenoids in a light- and temperature-dependent process. At 30°C the emission typically consumes 1–2% of assimilated carbon, but the portion can be considerably increased under drought and hot-weather conditions (Sharkey & Loreto, 1993; Bertin & Staudt, 1996; Staudt & Bertin, 1998). The physiological role of this type of VOC production is still an enigma. There is increasing evidence that it protects the photosynthetic apparatus against high-temperature damage by enhancing membrane stability, or against oxidative stresses by scavenging oxidants such as ozone (Loreto & Velikova, 2001; Sharkey et al., 2001; Peñuelas & Llusià, 2002). It may also act as a redox valve, preventing the phosphate limitations of photosynthesis (Logan et al., 2000).
The genus Quercus (oaks) includes > 300 woody species. It is widespread in the northern hemisphere, where it represents the dominant vegetation of temperate forests and shrublands (Nixon, 1993). Almost all the investigated species of oak were found to produce the hemiterpene isoprene, as many other terrestrial plants do. Interestingly, some Eurasian species, in particular Mediterranean oaks, show diversification in VOC production (Loreto et al., 1998; Csiky & Seufert, 1999). Quercus ilex (Holm oak) and Quercus suber (Cork oak), two widespread evergreen oaks in the western Mediterranean region, have been reported to be, respectively, a monoterpene emitter and a nonemitter instead of isoprene emitters (Kesselmeier & Staudt, 1999; Loreto, 2002). The assignment of Q. suber as a nonemitting species relies on data from measurements made on a few individuals in Italy (Steinbrecher et al., 1997; Delfine et al., 2000). Yet an earlier study in Portugal (Pio et al., 1993), as well as a recent study by Owen et al. (2001), report strong monoterpene emissions from this species. This discrepancy indicates the existence of geographic variability in the overall capacity to emit volatiles, which may be related to past genetic isolation of populations, adaptations to local growth conditions, or hybridization between emitting and nonemitting oak species. Quercus ilex and Q. suber are not very close genetically and belong to two distinct clades (Manos et al., 1999; Belahbib et al., 2001; Toumi & Lumaret, 2001; Bellarosa et al., 2004), for example (according to Schwarz, 1937) to the subsections Sclerophyllodrys (Schwarz) and Cerris (Spach), respectively. However, recent studies based on interspecific diagnostic genetic markers (mainly chloroplast DNA and allozymes) have pointed out nuclear and cytoplasmic genome exchanges caused by hybridization between Q. suber and Q. ilex in some geographic areas, particularly in the eastern and southern parts of Morocco and the eastern part of the Iberian Peninsula (Toumi & Lumaret, 1998; Belahbib et al., 2001; Lumaret et al., 2002; Castro et al., 2003).
There have been no attempts to investigate the intraspecific variability of emissions in Q. suber, as is the case for many other species. For Q. ilex some emission studies were conducted on populations of several dozen individuals: these were either adult trees, to explore emission variability in natural forest; or potted trees, to study emission responses to environmental conditions (Staudt et al., 2001a, 2001b, 2003). In all populations, distinct differences in emission patterns were observed among individuals. Emission signatures of individuals were highly preserved with respect to leaf age and leaf crown position, and were independent of treatments that changed emission rates by more than one order of magnitude. These results provide clear evidence of genetic variability at population level. Further investigation is needed to understand the extent and origins of isoprenoid emission diversification in Quercus and other taxa, and to improve the accuracy of large-scale emission inventories. Furthermore, it offers the opportunity to use emission signatures as chemotaxonomic markers to identify subspecies and to define provenances, as has been done with plant extracts in numerous monoterpene-storing plant species (Lewinsohn et al., 1993).
In the present study, we investigated for the first time the intraspecific variability of isoprenoid emissions from Q. suber, and expanded our previous study on Q. ilex to include more western and eastern populations in France (Staudt et al., 2001a). We characterized isoprenoid emissions of Cork oaks and Holm oaks growing in two mixed stands in sites located in two French regions which had been shown to differ substantially in their level of interspecific genetic exchanges between the two species. One region (Catalonia) was characterized by total, and probably ancient, cytoplasmic introgression of Q. suber by the Q. ilex genome and, on the basis of nuclear diagnostic markers, by a substantial proportion of individuals from more recent hybrid origin. The other region (Provence) showed a very reduced level of cytoplasmic interspecific exchange and limited nuclear gene flow occurring locally (Toumi & Lumaret, 1998; Lumaret, 2003; Lumaret et al., 2003). The main objectives are: to (i) analyse and compare intra- and interspecific quantitative and qualitative variation for VOC emissions; and (ii) identify the putative effect of genetic exchanges between the two oak species by assuming that higher interspecific exchanges may result in higher intraspecific variation for VOC production in each species.
Materials and Methods
A total of 90 trees were studied in two sites of southern France (Fig. 1). Site one, Mas Anglade, is located in French Catalonia in the eastern hills of the Pyrenean Mountains (42°29′19″ N, 02°51′39″ E, elevation 300 m) close to the Spanish border. In that region Quercus ilex L. can be considered the native and predominant tree species. However, many woodlands have been managed for cork production for at least two centuries. In the measuring site, approx. 80% of the trees are Quercus suber L., 15%Q. ilex and 5% other species such as the white oak Quercus pubescens Willd. Approximately two-thirds of the trees studied are located on the plateau and western slope of the site, whereas the other third are growing in a small valley. Twenty-one adult trees with the morphological appearance of Q. ilex (or close to it, see below) and 21 adult trees with the appearance of Q. suber were labelled and collected at that site.
Site two is located in Port-Cros island (43°00′06″ N, 06°22′36″ E, elevation 0 m), a national park in Provence region. Port-Cros is part of the Hyères islands which lie approx. 10 km offshore of the southernmost point of the French Riviera. In Port-Cros approximately half the trees were located inland as understorey trees in forests dominated by Aleppo pine (Pinus halepensis Mill.), and the other half in the coastal vegetation of the island's western seashore. At that site, which did not show any evidence of management, 25 adult trees morphologically looking Q. ilex, 22 adult trees with the morphological appearance of Q. suber, and one juvenile tree (approx. 10–15 yr old), showing leaf traits between evergreen and deciduous oaks, were analysed.
Genetic characterization of individual trees
In a first step, individuals were assigned to Q. ilex or Q. suber on the basis of several morphological traits according to species identification in Flora Europaea (Tutin et al., 1993), but primarily according to the cortical production of cork. However, as shown from experimental crosses, first-generation hybrid individuals have no cork (Natividade, 1936). Even for most other characters these individuals are usually closer morphologically to Q. ilex, and cannot be easily distinguished in natural conditions (Lumaret et al., 2002). Moreover, in natural interspecific hybridization, on the basis of variation in cytoplasmic DNA (cpDNA), which is maternally inherited in oaks (Dumolin et al., 1995), Q. ilex was found to be predominantly (but not exclusively) the mother species (Lumaret et al., 2002). Initial hybridization events were usually followed by backcrosses with either parental species, Q. suber being predominantly the pollen-bearing species. Consequently, cytoplasmic introgression was observed to occur mainly from Q. ilex into Q. suber and to be maintained over very long periods in large areas (Belahbib et al., 2001). Conversely, morphologically intermediate individuals between Q. ilex and Q. suber are rare, even in mixed stands. This suggests that the hybrids may have to cope with selective disadvantage, and that most of the initial unidirectional or bidirectional nuclear genetic introgression (including the genes coding for morphological traits) which may have occurred between the two oak species disappears with successive backcrosses, more particularly if these involve the same parental species. Therefore first-generation hybrids are expected to combine alleles specific to each parental species at each nuclear diagnostic locus, and the individuals, which correspond morphologically to one species but still possess nuclear alleles specific to the other, probably derive from a relatively recent interspecific hybridization event. Conversely, trees belonging morphologically to one species but which show the cpDNA molecular characteristics of the other parental species, without evidence of nuclear introgression, can be considered to derive from a more ancient hybridization event. In the present study, each individual tree collected for VOC emission analysis was characterized genetically for cytoplasmic and nuclear markers, namely restriction fragment-length polymorphism (RFLP) of cpDNA, which corresponds to very distinct molecular lineages in Q. ilex and Q. suber (Belahbib et al., 2001; Petit et al., 2003); allozyme variation at three nuclear loci located on distinct chromosomes and coding for phosphoglucose isomerases (EC 220.127.116.11), leucine aminopeptidases (EC 18.104.22.168) and acid phosphatases (EC 22.214.171.124). All these markers had been shown to be diagnostic between the two oak species, and the respective analytical techniques have been described previously (Toumi & Lumaret, 1998, 2001; Lumaret et al., 2002). In addition, using the same methodology and appropriate specific enzyme markers (Toumi & Lumaret, 2001), the putative hybrid origin was tested of a juvenile tree growing in Port-Cros and observed to be morphologically intermediate between an evergreen oak species and a deciduous oak species.
Measurements of isoprenoid emissions, leaf gas exchange and morphological traits
Measurements of isoprenoid emissions, leaf gas exchange (photosynthesis, transpiration), and structural traits were made on cut branches sampled during two campaigns, one in autumn (end October 2001, 85 trees) and one in summer (June 2002, 90 trees). Samples were usually taken from the upper sun-exposed parts of the tree crown by means of 9 m telescopic pole pruners. However, during winter 2001/02 three Q. ilex trees in the Catalonian site were chopped down by the farmers, so the summer measurements of these trees had to be made on leaves of resprouts. Weather conditions in the summer campaign were good and stable throughout all sampling days at both sites. In the autumn campaign they were variable: During sampling days weather conditions were balmy and sunny at the Catalonian site, while cool and stormy on the island of Port-Cros.
Sampled branches were immediately submerged in water and re-cut underwater to avoid xylem air embolism. They were transported in water to the Institute and stored either in a glasshouse (autumn campaign) or outdoors in a half-shaded place (summer campaign) until all samples were assayed for isoprenoid emissions. Emission assays lasted for 3–9 d, during which water was regularly replaced. After measurements, leaves were harvested to determine their fresh weight (microbalance Mettler PM200, Mettler-Toledo SA, Virofley, France), projected area (Delta-T Area Meter MK2, Delta-T Devices Ltd, Cambridge, UK), and dry weight after drying for at least 48 h in a ventilated oven at 60°C. These data were used to calculate specific leaf mass (SLM, leaf dry mass per unit projected leaf area in g m−2) and relative leaf water content (LWC, %) here defined as: (fresh weight − dry weight)/fresh weight × 100.
Foliar isoprenoid emissions and gas exchange were determined indoors by clamping one or several terminal leaves of a sample in a temperature and light-controlled dynamic flowthrough gas-exchange chamber located in a nearby laboratory. The custom-made chamber (volume 100 ml) consists of a transparent lid and a water jacket and Teflon-coated metal housing equipped with two thermocouples to measure leaf and air temperature, and a quantum sensor (Licor, PAR-SB 190, Lincoln, NE, USA) to measure photosynthetic photon flux density (PPFD). Leaf gas exchange (photosynthesis, transpiration) was measured by drawing a constant portion of the inlet and outlet air through a CI-301 infrared CO2 gas analyser (CID Inc., Camas, WA, USA) via tubes enclosing two humidity sensors (HIH-3602C, Honeywell Inc., IL, USA). Data were collected by a 21X datalogger (Campbell Scientific Ltd, Shepsherd, UK). The net airflow rate inside the chamber was 0.32 mmol s−1. Measurements were made at a constant leaf temperature of 30 ± 1°C and PPFD of approx. 900 µmol m−2 s−1 after an acclimation time of at least 1 h under the same conditions. The humidity of the inlet air was set to approx. 50% by bypassing a portion of the air stream through a humidifier.
Several trees were repeatedly assayed in a campaign because leaves of their samples showed low physiological activity. When photosynthesis was lower than approx. 3 µmol m−2 s−1, leaves of the same branch were re-assayed during the following days, and in case activity remained low a new branch was sampled and re-assayed under the same conditions. Generally, in summer almost all branches sampled showed immediate normal photosynthetic activity, whereas in autumn many samples, in particular those of Q. suber, became active only after some days. Samples of some trees in the autumn remained completely inactive. The data for these measurements were not considered in the data evaluation because they do not allow conclusions on the tree's capacity to emit isoprenoids.
Isoprenoid emissions were measured by drawing cuvette exhaust air (1 l at 0.1 l min−1) through a glass tube containing 200 mg of either Tenax, for monoterpenes and higher terpenes or Carbotrap, for isoprene. These traps were analysed by capillary gas chromatography (Chrompack CP9003 with flame ionization detector, Varian SA, Les Ulis, France) after thermal desorption (10 min at 200°C) and preconcentration on a capillary cold trap (−100°C for monoterpenes, −150°C for isoprene) by a TCT 4002 thermodesorber (Chrompack). Complete resolution of the monoterpenes emitted was achieved with a Sil 8CB fused silica capillary column (25 m × 0.32 mm, 1.2 µm df, Chrompack) using the following multiramp temperature program: 4 min at 65°C, 2.5°C min−1 to 80°C, 2°C min−1 to 100°C, 10°C min−1 to 220°C. Isoprene emission was analysed according to Van Eijk & Kotzias (1994) using a fused silica PLOT Al2O3/KCl column (25 m × 0.32 mm, Chrompack). The temperature program was 2 min at 80°C, 8°C min−1 to 200°C, 10 min at 200°C. Carrier gas was helium (1.2 ml min−1) in all cases. Both precision and accuracy were within 2–5% as determined by repeated measurements of standards (Fluka Chemie AG, Buchs, Switzerland; Roth, Karlsruhe, Germany) at realistic concentrations. The lower detection limit of the analytical system is approx. 0.5 p.p.b. (v/v) for the monoterpenes.
Data for the summer and autumn campaigns were treated separately. For each data set, one-way and two-way anovas (sigmastat 3.0, SPSS Inc., Chicago, IL, USA) were used to test differences among genotypes, species and geographical origin (site), and interactions. F tests were used to determine the statistical significance (P < 0.05), and the Student–Newman–Keuls test for the multiple pair-wise comparisons. Kruskal–Wallis anovas were applied to ranked data if tests for normality or equal variance failed.
In a first step, we tested whether Q. ilex and Q. suber morphotypes, identified by nuclear genetic markers to be of recent hybrid origin between Q. ilex and Q. suber (see below), differed from ‘pure’ individuals (trees with no evidence of recent hybrid origin) in their leaf emission, gas exchange or anatomical features. In both sites and for both campaigns, statistically significant differences were seen only between the two species (individuals from recent hybridization or not), but not between ‘pure’ individuals and individuals from recent hybrid origin within a species (one-way anova, P > 0.05). Because of the absence of significant differences in emission and of measured variables, data for individuals derived from recent hybrids were pooled to the respective data sets for the ‘pure’ trees and analysed for differences with respect to the factors species and site (two-way anovas).
Additional t tests were applied to evaluate the effects of growth conditions (shaded vs open vegetation) within a species and site, and differences between emitting and nonemitting Q. suber trees. Pearson's product moment correlation or, in case residuals showed no normal distribution (determined by the Kolmogorov–Smirnov test), Spearman's rank order correlation (all sigmastat 3.0) were used to detect correlations between measured variables. Cluster analysis was carried out on the fraction of individual monoterpenes using Ward's technique with the Euclidian distance measure (R-GUI Software, GNU Operating System Free Software Foundation, http://www.gnu.org/). χ2 analysis of contingency tables (sigmastat 3.0) was used to compare frequencies of emission types (main clusters) between populations.
Individual genetic characterization
All Q. suber and Holm individuals from Catalonia (Mas Anglade) possessed a cpDNA haplotype belonging to the Q. ilex lineage, indicating a complete cytoplasmic introgression of Q. suber by Q. ilex in that site (Table 1). In the Provence site (Port-Cros), all trees analysed possessed a cpDNA haplotype corresponding to their respective morphology, except one Q. ilex tree and one Q. suber tree, for which the cpDNA of the opposite species was observed. On the basis of nuclear genetic markers, 19 trees, seven in Provence (15% of the trees sampled in that site) and 12 in Catalonia (29%) combined alleles of both species at least at one locus, and were therefore identified as of recent hybrid origin (Table 1). Among these, most individuals in the Provence site possessed alleles of the opposite species at two or three loci, whereas most individuals in the Catalonian site did so only at a single locus. This indicates that the hybridization events from which these individuals derived are more recent in the Provence site than in the Catalonian site.
Table 1. Number of trees from Port-Cros (Provence) and Mas Anglade (Catalonia) showing either a Quercus ilex or a Quercus suber morphotype and possessing a cpDNA molecule and/or alleles at one, two or three diagnostic allozyme loci characteristic of the opposite species
Q. ilex (25)
Q. suber (22)
Q. ilex (21)
Q. suber (21)
Generally, most individuals of recent hybrid origin had leaf morphology close to Q. ilex, and no cortical cork. One apparently different tree on Port-Cros was identified as a hybrid between Q. ilex and Quercus canariensis Willd., an isoprene-emitting deciduous oak species introduced in Port-Cros approx. 60 yr ago (Roux & Thinon, 1987). On the basis of cpDNA variation analysed in that hybrid, Q. ilex was identified as the mother parental species.
Intra- and interspecific emission variability
None of the Q. suber and Q. ilex trees studied emitted significant amounts of isoprene, except the Q. ilex × Q. canariensis hybrid which emitted isoprene and monoterpenes in proportions of approx. 60/40% (Fig. 2). Four Q. suber trees in the Catalonian site did not emit isoprene or monoterpenes in significant amounts, although their leaves were physiologically fully active during both campaigns (Fig. 2, trees on extreme right). These trees were considered as nonemitters. Foliage of all other trees emitted the same principal monoterpenes: α-pinene, β-pinene, sabinene, myrcene and limonene (Fig. 2b). In both species and sites, the fraction of individual compounds (percentage of total emission) varied among individual trees. For a given tree the emission profile was almost stable. Commonly the limonene portions decreased by 2–9% from the autumn to the summer campaign, while the percentage of other main compounds slightly increased. Several other monoterpenes were found in traces but were close to the detection limit and therefore discarded from further data evaluation.
Mean emission rates of the sum of the five main compounds determined at standard conditions were significantly different between Q. ilex and Q. suber in both sites and during both campaigns. In autumn the emission rates of Q. ilex in the Provence and Catalonia sites were 4.7 and 2.8 times higher, respectively, than those of Q. suber collected at the same sites (Table 2). In autumn there was also a significant difference between the Provence and Catalonia sites: the emission rates of Provence samples collected at the beginning of November, during a cold and rainy period, were substantially lower than those of the Catalonia samples, collected at end of October under mild, sunny weather conditions (Table 2). The lowered emission rates of the Port-Cros samples were also paralleled by lowered rates of photosynthesis, although differences in photosynthesis were not significant. In the summer campaign weather conditions were stable and no site differences were seen in emission or photosynthesis rates. Standard emission rates were much higher than in the autumn, especially those of Q. suber. Unlike what was observed in the autumn, in summer leaves of Q. suber emitted significantly more monoterpenes than Q. ilex leaves on a leaf-area basis, and particularly on a leaf dry-matter basis.
Table 2. Leaf monoterpene emissions, photosynthesis and anatomical features of Quercus ilex and Quercus suber trees growing in mixed stands in two sites, one in Provence (Port-Cros) and one in French Catalonia (Mas Anglade)
Site and species
Emission composition (sum = 100%)
Photosynth (µmol m−2 s−1)
SLM (g m−2)
Leaf form (length/width)
nmol m−2 s−1
µg g−1 h−1
On each tree, emission and photosynthesis were measured in autumn and summer on current-year leaves at the same temperature and light conditions of 30°C and 900 µmol m−2 s−1 PPFD. SLM, specific leaf weight; LWC, relative water content of leaves. Values are means ± SE. P values (two-way anova) indicate significant differences between site, species and interactions (P < 0.05).
Only data from physiologically active and emitting Q. suber and Q. ilex trees were included (see Materials and Methods).
The summer campaign in the Catalonian site includes data from resprouts of three Q. ilex trees (see Materials and Methods) showing extremely low SLM (130 g m−2). Excluding these data, mean SLM is 215 gm−2 and site differences are significant (P = 0.002).
In both campaigns mean fractions of four monoterpenes were significantly different between the two species (Table 2), except the limonene fractions which showed strong intraspecific variations. Differences in mean fractions were more pronounced in the Provence site than in the Catalonian site, and a site effect was detected for the α-pinene, sabinene and myrcene fractions in emissions from Q. ilex during summer (species–site interaction, P < 0.05). Cluster analysis of the emission signatures (Fig. 3) split trees into four main types, frequencies of which differ significantly among populations in some cases (Table 3). First, the pinene type which emitted mainly α-, β-pinene plus sabinene – this was the most frequent type (72% of Q. ilex and 51% of Q. suber) and was particularly common in the Q. ilex population at the Provence site (80% of trees). Second, a rather rare type emitting high limonene was observed exclusively in the Q. suber population of the Provence site, with exception of one Q. ilex tree derived from recent interspecific hybridization at the Catalonian site. Third, a more common limonene type which emitted limonene in intermediate proportions – this type was present in all populations but was more frequent in the Catalonian Q. suber population (53% vs 20–28% in the other populations). The final type emitted myrcene in high proportions and was represented by a sole Q. ilex tree at the Catalonian site. Within the main clusters, several subclusters appeared that separate the two oak species well, except for two and four trees in the pinene and limonene clusters, respectively (Fig. 3). More particularly, in the pinene-type cluster one Q. ilex and one Q. suber, both from Provence, were classified with trees of the opposite species. In the limonene-type clusters a Catalonian tree that showed Q. ilex morphology (or very close to it) and possessed one and two alleles peculiar to Q. suber at two and one diagnostic enzyme loci, respectively (i.e. not a first-generation hybrid but an individual probably derived from recent interspecific hybridization), was classified in the subgroup of Provencal trees producing a high proportion of limonene. The third exception concerned three trees from Provence with the morphological appearance of Q. ilex (one of recent hybrid origin and two with no evidence of hybrid origin) that were classified with Q. suber trees producing intermediate limonene proportion. Generally, discrimination of the two species by emission signature was largely caused by differences in the fractions of sabinene and myrcene in relation to other compounds in the emissions, Q. suber leaves emitting relatively more sabinene and less myrcene than Q. ilex leaves. For instance, the ratio of sabinene to α-pinene in the emissions of Q. suber was higher and more stable throughout the emission types than that of Q. ilex, and this allows fairly good distinction of the two species at both sites (Fig. 4).
Table 3. Absolute and relative frequency of four emission types within two sites of mixed Quercus ilex and Quercus suber populations in southern France (cf. Fig. 1)
Site/species (number of trees studied)
Quercus suber and Q. ilex trees were assigned to emission type according the dendrogram in Fig. 3. Frequencies of emission type in sites and species were compared using χ2 analysis of contingency tables. *The four nonemitting Q. suber trees at the Catalonian site are not considered.
Distinct correlation patterns among monoterpene fractions were found in the emissions of Q. suber, revealing two classes of compound: one including α-pinene, β-pinene and sabinene; the other myrcene and limonene (data not shown). In each class, each compound was strongly positively correlated to the other compound(s) of the same class (r = +0.88 to +0.98, P < 0.0001) and was strongly negatively correlated to each compounds of the other class (r = −0.86 to −0.99, P < 0.0001). In the emissions of Q. ilex, monoterpene fractions were rather scattered: as in Q. suber, α-pinene, β-pinene and sabinene scaled positively to each other but correlations were less tight than in Q. suber (r = +0.35 to +0.87, P = 0.0168 to < 0.0001). However, a strong positive correlation was seen between α-pinene and the sum of β-pinene and sabinene (r = +0.95, P < 0.0001). Limonene fractions showed a negative correlation to α-pinene and sabinene (r = −0.63 to −0.77, P < 0.0001) and less accurately to β-pinene fractions (r = −0.36, P = 0.0142). No correlations were seen between myrcene and α-pinene, sabinene or limonene fractions (P = 0.05). Myrcene fractions were negatively correlated to β-pinene fractions (r = −0.56, P < 0.0001).
Leaf morphological traits
The tree's emission characteristics in terms of quantity and quality were also compared with leaf morphological traits and gas-exchange data. In Q. ilex standard emission rates were positively correlated to LWC (r = +0.49, P = 0.0005) and to photosynthesis rates (r = +0.54, P = 0.0001), and slightly negatively to SLM (r = −0.39, P = 0.0070) (data not shown). No corresponding correlations were seen in Q. suber. In both species there was no correlation between emission type and total emission rate or leaf morphological traits. Generally, at both sites the two species displayed significant differences in average leaf form, SLM and LWC (Table 2): Q. suber had more round leaves with lower SLM and higher LWC. The SLM scaled negatively with LWC (r = −0.62, P < 0.0001) and leaf size (r = −0.34, P = 0.0015) across species, sites and seasons (data not shown), suggesting that variation in SLM depends not only on variation in leaf thickness, but also on the concentration of leaf dry matter. In both species SLM was higher in autumn than in summer and, inversely, LWC was lower in autumn than in summer (Table 2), which indicates that leaf dry matter concentration should have increased with leaf age over the seasons. Between the two sites, SLM was lower and LWC higher in the Provence than in the Catalonian population (Table 2). Leaves from trees of the Provence site also tended to be more elongated and to be larger. These site differences in anatomical traits may reflect the specific growth conditions of the trees’ habitats: on one hand the rather dense and shaded environment of the forests on the Port-Cros island, and on the other the open oak orchard in the Catalonian hills.
Quercus suber has been classified as a nonemitting oak species on the basis of a few measurements made in Italy (Steinbrecher et al., 1997). In contrast, in the present study almost all Q. suber trees (91%) were found to be monoterpene emitters. In summer their mean standard emission rate was 41.5 µg g−1 h−1 (13.8 nmol m−2 s−1), which significantly exceeded that of cohabiting Q. ilex trees (26.1 µg g−1 h−1, 10.2 nmol m−2 s−1). This basal emission rate is among the highest ever reported for monoterpene-emitting plants (cf. Kesselmeier & Staudt, 1999). Consequently the regional emission inventories that computed Q. suber forests as a nonemitting vegetation may have largely underestimated the annual release of reactive VOCs in the Mediterranean basin, and therefore should be revised. The basal emission rates of Q. ilex measured in summer and autumn on cut branches are consistent with those reported in previous studies for intact branches (Bertin et al., 1997; Sabillón & Cremades, 2001; Staudt et al., 2002; Rapparini et al., 2004), which validates the method used in our study. Basal emission rates of Q. ilex and Q. suber decreased strongly and were site-dependent in the autumn campaign. Both seasonal and site effects can largely be attributed to weather conditions which affect the tree's capacity to produce monoterpenes. It has been demonstrated in a recent growth-chamber study that the basal emission rate of Q. ilex can be reversibly down- and upregulated with prevailing temperature and light conditions (Staudt et al., 2003). However, among the two species, autumn emissions were much more reduced in Q. suber than in Q. ilex compared with their emission rates in summer. They were about halved in Q. ilex, but reduced more than sixfold in Q. suber. This greater seasonal drop in Q. suber leaves may be related to the shorter life span of Q. suber leaves. They became usually senescent after 1 yr during the new flush so that Q. suber canopies consisted mainly of only one leaf age class. In contrast, in Q. ilex 1- and 2-yr-old leaves were regularly found together with current-year leaves. Fischbach et al. (2002) found that the activity of monoterpene-producing enzymes in young Q. ilex leaves strongly decreases during autumn and winter, but partly recovers during the spring and summer in the second year. Staudt et al. (2003) observed a decrease in the foliar emission capacity of Q. ilex with leaf ageing and a fast drop during leaf senescence.
Standard emission rates of the two Q. suber populations were almost equal (Table 2), despite their genetic differences in terms of provenances (Lumaret, 2003) and level of introgression by Q. ilex. This result, and the fact that nonemitting Cork oaks were not found in the site characterized by low interspecific genetic flow, provide evidence that the observed capacity to produce monoterpenes in Q. suber has not been acquired through local genetic introgression by Q. ilex, but is an intrinsic property of Q. suber. It is possible that the capacity to produce monoterpenes has been secondarily gained or lost in some Q. suber provenances. A screening study carried out in Portugal by Pio et al. (1993) described emission data for Q. suber that match our results in both quantity and quality. The Iberian Peninsula has been recognized as a centre of high genetic diversity of Q. suber and may constitute a putative centre of origin and diversification (Toumi & Lumaret, 1998). More probably it may be a secondary centre, as suggested by Bellarosa et al. (2004) on the basis of rDNA variation, and the most ancient forms would be located in south-east continental Italy. In a very recent study of cpDNA variation over the whole geographic distribution of Q. suber, three putative glaciary refugia were identified in southern Italy and North Africa, from which the Provencal populations are derived, and in the south of the Iberian Peninsula, probably the origin of the Catalonian populations, respectively (R.L. and co-workers, unpublished data). Thus, assuming that the Italian (most ancient) Q. suber provenances are nonemitters, as indicated by the measurements of Steinbrecher et al. (1997), the gain may be caused by disjunctions in refugia during glaciations, and to subsequent recolonizations. Alternatively, the losses or gains in emission capacity may have occurred many times in many populations, as indicated by the presence of some nonemitting individuals among emitting Q. suber trees, which was never observed in Q. ilex populations (Staudt et al., 2001a), nor in any other isoprenoid-emitting plant species.
While standard emission rates largely changed from autumn to summer, the emission signature of a given tree was quite stable in Q. suber as well as in Q. ilex, as observed in many previous studies (Bertin et al., 1997; Sabillón & Cremades, 2001; Staudt et al., 2001a, 2001b, 2002, 2003; Rapparini et al., 2004). Hence in both species genetic variability in monoterpene synthesis exists that largely accounts for the observed variability in the fractions of the five major monoterpenes emitted from individuals. Monoterpenes are synthesized by a class of enzymes called monoterpene synthases, of which some produce a sole monoterpene, and others several monoterpenes in defined proportions (Trapp & Croteau, 2001). They are encoded by nuclear genes, which is consistent with the observation that hybrids of isoprene- and monoterpene-emitting species, here Q. ilex and Q. canariensis, emit both isoprene and monoterpenes (Schnitzler et al., 2004), because chloroplast genes are only maternally inherited in oaks (Dumolin et al., 1995). Within the studied Q. ilex and Q. suber populations the proportion of emitted monoterpenes did not vary randomly, but showed some distinct chemotypes (Figs 2, 3) independently of the total amount of monoterpenes emitted. This finding, and the correlation pattern observed among monoterpene fractions in the emission signatures, evoke the occurrence of coordinated synthesis of some terpenes by genotype-dependent monoterpene synthases. In the case of Q. suber, the tight positive correlations between α-pinene, β-pinene, sabinene on one hand and between limonene and myrcene on the other, and the tight negative correlations between compounds of these two groups, suggest that basically two enzymes exist which, depending on their dominance, would generate pinene emitters, limonene emitters or intermediate limonene/pinene emitters. In Q. ilex populations several other monoterpene synthases may exist, because emission signatures were more variable and their monoterpene fractions less correlated one to each other than those in Q. suber populations: We speculate that α-, β-pinenes plus sabinene are produced by at least two different enzymes in Q. ilex, one producing more β-pinene and the other more sabinene. This would explain why α-pinene fractions were very well correlated to the sum of β-pinene and sabinene fractions, but less to the fractions of each single compound. Limonene, when emitted in large portions, should be mainly produced by a rather specific enzyme, as in Q. suber. Finally, myrcene, which in Q. ilex is present in greater proportion than in Q. suber, is probably a by-product of several other enzymes and, in addition, is synthesized by a specific monoterpene synthase producing predominantly this compound. The gene encoding such an enzyme has recently been isolated from Q. ilex leaves by Fischbach et al. (2001). Apparently this gene is rarely expressed in high activity. In the present study only one tree was found to release myrcene in high fractions, which represents 2.4% of the Q. ilex trees investigated (Table 3). In a previous study carried out in the Languedoc region, 8% of Q. ilex trees were identified as strong myrcene emitters (Staudt et al., 2001a) and 21% as limonene emitters (high and intermediate types together). This frequency of limonene emitters is close to that observed in the Provence site (20%) but lower than that found in the Catalonian site (33%). It has been suggested that the limonene-emission type may be more typical of western Q. ilex provenances, especially of the Q. ilex subspecies rotundifolia (Csiky & Seufert, 1999), a Q. ilex morphotype characterized by round, small leaves which grows in Iberia and North Africa. Our data do not support such a relationship, because the emission signatures of Q. ilex trees were not associated with specific leaf form. Trees expressing a limonene-emission type were found throughout all populations across species, growth conditions and leaf morphology (Staudt et al., 2001a, 2001b, 2003).
Interspecific gene flow may diversify the emission signatures of oaks or give rise to new chemotypes, as illustrated by the peculiar emission profile of the hybrid Q. ilex × Q. canariensis. Unlike the complete loss of monoterpene production, purely qualitative alterations in emission pattern should not affect the ability of individuals to compete, and therefore may be conserved in populations. Emission signatures of individuals derived from hybridization between Q. ilex and Q. suber showed a large range of chemotypes and, on average, were not significantly different from the signatures emitted by the other individuals. Yet some individuals derived from recent hybridization possessed a rare chemotype, e.g. one tree characterized by high limonene production, which was not observed in any of the other oak trees analysed in the same site, whatever the species. Moreover, as mentioned above, a site effect was indicated for the two Q. ilex populations, which differed in their abundance of individuals of recent hybrid origin (38% in the Catalonian vs 20% in the Provence site), and mean emissions of which displayed significant differences in the sabinene, myrcene and α-pinene fractions (Table 2). Therefore hybridization events in Q. suber and Q. ilex populations probably contribute to diversity in emission signatures, but this could hardly be detected from the low number of observations because the emission composition of the two species are very close and show similar intraspecific variations. Nevertheless, significant differences between Q. suber and Q. ilex were seen in the mean fractions of most of the emitted monoterpenes (Table 2). Generally, in all chemotypes Q. suber emitted higher proportions of sabinene and lower proportions of myrcene than Q. ilex, which may be of help in identifying the species and their hybrids. In a first estimate, a good diagnostic trait appears to be the sabinene to α-pinene ratio, with a threshold around 0.75 for the populations investigated (Fig. 4). However, the emissions of several trees had ratios close to this threshold and hence would not be easily identified by this trait. These include several individuals that originated from recent interspecific hybridization, and others for which no evidence of recent hybrid origin could be obtained, which may also be caused by the limited number of nuclear diagnostic markers used in the study. In addition, as in crosses between these two species, Q. ilex is usually the mother species; there is a low probability of identifying a Q. suber cpDNA molecule in the hybrids.
Isoprenoid emissions from oaks have been proposed as chemotaxonomic markers to differentiate two groups, Sclerophyllodrys (monoterpene emitters) and Cerris (nonemitters), from other oak clades (all isoprene emitters) (Loreto et al., 1998; Csiky & Seufert, 1999). The Sclerophyllodrys group consists of a few Mediterranean oak species (Q. ilex sensu lato and Q. coccifera s.l.), whereas the Cerris group embraces a mixture of European and Asian oaks including Q. suber (Loreto, 2002). Our results confirm that monoterpene emission is an attribute of Q. ilex and hence of the Sclerophyllodrys group, but they cast doubt on whether nonemission is a reliable marker of the Cerris group. According to the data compiled by Loreto (2002), already two of the nine oak species investigated in the Cerris group, Quercus ithaburiensis and Quercus semecarpifolia, do not share this attribute, as they have been identified as monoterpene and isoprene emitters, respectively. If these two oaks are correctly classified in the Cerris group, and if Q. suber is intrinsically a monoterpene-emitting species, as suggested by our data, then the Cerris group is not well discriminated from other oaks in terms of isoprenoid emissions. However, no reliable conclusions can be drawn as long as species are assigned to emission types on the basis of a few measurements only. As highlighted by our study on Q. suber, important variability exists within and among populations of the same species that merits further investigation.
We gratefully acknowledge support by the staff of the national park of Port-Cros and, in particular, the indispensable help of Patrice D’Onofrio. We thank François Jardon and Violette Sarda for their assistance in collecting plant material, Alain Rocheteau for helping to install the enclosure system, Thomas Verron for running the cluster analysis, and Jean Pierre Ratte for providing the map. This study received financial aid from the GIP ECOFOR project, the European Associated Laboratory of the CNRS and the French Ministry of the Environment.