Competitive strategies for water availability in two Mediterranean Quercus species

Authors


A. Nardini Fax: + 39 40 568855; e-mail: salleo@univ.trieste.it

Abstract

Competition for water availability was studied in a mixed natural stand of Quercus suber L. and Quercus cerris L. growing in Sicily by measuring diurnal changes of leaf conductance to water vapour (gL), water potential (ΨL) and relative water content (RWC) in April, July and October 1997 as well as the seasonal changes in root hydraulic conductance per unit leaf surface area (KRL). Quercus cerris behaved as a drought-tolerant species, with strong reductions of KRL, ΨL, and RWC in the summer. By contrast, Q. suber appeared to withstand summer drought by an avoidance strategy based on reducing gL, maintaining ΨL and RWC high and KRL at the same level as that measured in the spring. A ‘conductance ratio’ (CR) was calculated in terms of the ratio of gL to KRL. Seasonal changes of this ratio contrasted in the two species, thus suggesting that Q. suber and Q. cerris did not really compete for available water. In the summer, when Q. suber was extracting water from the soil to maintain high leaf hydration, Q. cerris had restricted water absorption, thus suffering drought but tolerating its effects. The possibility that cohabitation of drought-tolerant with drought-avoiding species can be generalized is also discussed.

INTRODUCTION

The seasonal and diurnal dynamics of parameters describing the water relations of plants, measured in different plant organs, have allowed us to define some of the main strategies adopted by plants to withstand drought stress (e.g. Levitt 1980; Nilsen & Orcutt 1996). In this regard, one of the most interesting aspects of adaptive processes is the competition for available water among species growing in the same dry areas.

As an example, wild olive tree (Olea oleaster Hoffmgg. et Link) and carob tree (Ceratonia siliqua L.) are typical components of the vegetation growing in dry regions of the Mediterranean Basin. Both species have been shown to undergo substantial water losses during periods of summer water shortage (Lo Gullo & Salleo 1988) but C. siliqua was able to compensate for water loss so that the leaves maintained relative water contents as high as 95%. On the contrary, the leaves of wild olive tree lost more than 25% of their water content and reached the turgor loss point. The effects of such pronounced leaf dehydration, however, were well tolerated by this species although at the cost of a much lower growth rate than that of the carob tree. In other words, carob trees were more successful in the competition for water thus avoiding drought whereas a strategy for drought resistance based on tolerance was a necessity for wild olive trees.

The evaluation of strategies for drought resistance in plants has been classically based on the dynamics of water relations parameters measured in leaves and stems, such as leaf water status (Lo Gullo & Salleo 1988; Nardini, Lo Gullo & Tracanelli 1996; Rico et al. 1996), evaporative flux (Reich & Hinckley 1989; Tsuda & Tyree 1997) and, more recently, stem and leaf vulnerability to xylem cavitation (Harvey & van den Driessche 1997; Kikuta et al. 1997; Tyree, Davis & Cochard 1994).

Nonetheless, it is in the soil that plants are closest to each other and competition for water is strongest. Therefore, an increasing number of studies of the water balance of plants have included measurements of the hydraulic behaviour of roots as a major factor determining the response of plants to different environmental stresses, including drought (e.g. Cochard, Ewers & Tyree 1994; Tyree et al. 1995; Sperry & Ikeda 1997; Steudle & Heydt 1997).

A better understanding of competition among different species for available water is of great importance not only from a theoretical point of view but also for practical purposes such as, for instance, reforestation. In fact, modern methods of reforestation now tend to exclude monocultural forests in favour of mixed forests which are considered more similar to natural vegetation and less vulnerable to summer fires (Ackzell 1996; Kräuchi & Xu 1996) and drought.

In an attempt to investigate more extensively the nature of the competitive strategies for drought resistance in forest species growing in Mediterranean areas, zones were selected in Sicily (southern Mediterranean Basin) characterized by mixed semi-natural forests. In particular, an area was selected (see below) consisting of mixed populations of Quercus suber L. (Cork oak) and Quercus cerris L. (Turkey oak). These species are known for growing in Sicily within quite distinct altitudinal ranges, i.e. between sea level and about 600 m altitude for Q. suber but up to over 700 m altitude for Q. cerris (Pignatti 1982). Therefore, the two oaks are likely to be subjected to different degrees of water stress in their natural environments and to resist drought to different extents. In particular, the stems of Q. cerris have been shown to be very vulnerable to xylem cavitation at leaf water potentials corresponding to the turgor loss point (Lo Gullo et al. 1995). Therefore, cohabitation of the two oaks on the same site was considered to be unusual and strong competition for available water in the soil was expected.

The present study deals with field measurements of water relations parameters measured in leaves and roots of Q. suber and Q. cerris on a seasonal time scale, with the purpose of: (a) defining their strategy for drought resistance and (b) investigating the nature of their competitive strategies for water absorption during natural periods of water shortage.

MATERIALS AND METHODS

Studies were made on plants of Quercus suber L. (an evergreen sclerophyll) and Quercus cerris L. (a deciduous tree), of 5–8 years of age.

The plants under study were the result of assisted natural regeneration of an adult forest (see below). Their height (h), trunk diameter (ØT) and total leaf surface area (AL) are reported in Table 1. All measurements were made on two days of April (15 and 16), July (8 and 9) and October (14 and 15) 1997, corresponding to the end of the rainy season (April), to the warmest month in Sicily (July) and to the beginning of the autumn rainfall (October). In the case of Q. cerris, April coincided with bud burst and October was the period preceding winter dormancy, when the leaves showed the initial symptoms of senescence, i.e. the leaf edges were yellowish. In April, only about 20% of Q. suber twigs were growing while the shedding of old leaves had already occurred. In October over 80% of the twigs of Q. suber were growing actively.

Table 1.  . Dimensions of young plants of Quercus suber and Quercus cerris, in terms of height (h), trunk diameter (ØT) and total leaf surface area (one side only, AL). Means are given ± 1SD (n = 20) Thumbnail image of

Site description

The experimental site was selected with the assistance of the Regional Forestry Service of Sicily and was located at Piano Fico within the Mount Nebrodi Natural Park (Caronia, Northern Sicily), at 650 m altitude and with NE exposure. The soils in the area have been classified as a ‘Flysch di Reitano’. Such sedimentary soils consist of clastic deposits with intercalated sandy, conglomeratic and sometimes marly bodies. Mean annual rainfall in the area is about 365 mm, only 8–15% of which falls between the end of April and October. Mean annual temperatures vary between 3 and 5 °C in the winter (with occasional frosts) and 28 °C in summer. In the period of study, the diurnal temperatures were 14–20 °C in April, 22–28 °C in July and 15–20 °C in October and the relative humidity of the air was 45–55%.

The forest vegetation consisted of a homogeneously mixed stand of Q. suber and Q. cerris. Most trees of the two species were over 30 years old with no natural regeneration, except for some areas 50 m wide and over 300 m long where adult trees had been felled at the end of the 1980s with the purpose of controlling summer fires. Here, impressive natural regeneration of the mixed forest was observed with a plant density of 0·5–1 plant per square metre. All the new plants in this area were derived from seedlings. Measurements were made on these young trees because competition for water available in the soil is likely to have a strong impact on the water balance of plants during their juvenile phase.

Measurements of soil moisture content

Seasonal changes of the soil water content were estimated in terms of changes in the dielectric constant (Ka) of the soil, using a Trase System (6050XI; Soilmoisture Equipment Co., Santa Barbara, CA, USA). The instrument was equipped with dedicated software relating Ka to the soil moisture volumetric content. Measurements were made at 0·15, 0·30, 0·45 and 0·60 m depth. Deeper horizons could not be explored because of the rocks below.

Measurements of the leaf water status

The diurnal time course of leaf conductance to water vapour (gL), water potential (ΨL) and relative water content (RWC) was measured every 60 min between 0730 and 1830 h, in the three months mentioned above. All the measurements were made on leaves of the current year .

In particular, gL was measured on at least ten leaves per species and per daytime, while still attached to the trees, using a portable steady-state porometer (LI-1600; LiCor Inc., Lincoln, NE, USA). Each measurement was completed within about 30 s while maintaining the relative humidity of the air inside the porometer chamber near the ambient value. Ambient temperature and relative humidity were also recorded using the porometer cuvette held at about 1 m from the canopy.

Leaf water potential (ΨL) was measured on six to ten leaves per species and per daytime, on different trees using a portable pressure chamber (PMS 1000; PMS Instrument Company, Corvallis, OR, USA) with sheets of wet filter paper inside the chamber to minimize water loss during measurements.

The leaf RWC of at least 20 leaves per species and per daytime was measured, from different trees. The leaves were cut off while within plastic bags, and immediately weighed on a portable digital balance (MR50; Chyo Balance Corp., Kyoto, Japan) maintained in the shade. After their fresh mass (MF) was recorded, the leaves were enclosed in plastic sheets and placed in zip-lock plastic bags and kept in a thermal bag at about 4 °C. At the end of the experiments, the leaves were brought to the laboratory and resaturated with water to full turgor. This was achieved by immersing their petioles in distilled water, while covering the leaf blades with plastic bags and leaving the leaves in the dark overnight. The value of ΨL was remeasured to check that it was higher than –0·05 MPa with no leaf oversaturation. The leaves were reweighed to get their turgid mass (MT) and then dried at 70 °C for 3 days to get their dry mass (MD). The RWC was calculated as 100× (MFMD)/(MTMD).

In order to estimate changes in the leaf water status as a function of dehydration, pressure–volume (PV) curves (Salleo 1983; Tyree & Hammel 1972) of five to seven leaves of the two oak species were measured during each of the study months. From the PV curves, the leaf water potential at the turgor loss point (ΨTLP) and osmotic potential at full turgor (Ψπο) were measured and the bulk modulus of elasticity at full turgor (ɛmax) was calculated. In particular, ΨTLP served as a reference point for estimating the residual turgor of leaves when reaching the minimum diurnal ΨL while Ψπο could provide information on the eventual osmoregulatory processes during increasing drought. Finally, ɛmax was informative with respect to the potential ΨL drop in response to small symplastic water losses (Salleo & Lo Gullo 1990; Tyree & Karamanos 1980).

Measurements of root hydraulic conductance

At least three methods are reported in the literature for measuring the hydraulic conductance of root systems (KR) growing in the soil, i.e. the pressure chamber (Fiscus 1975; Rüdinger et al. 1994), the root pressure probe (Steudle 1993) and the high pressure flow meter (HPFM, Tyree et al. 1994, 1995). Although the most informative method is the root pressure probe, the HPFM is the only method developed so far (together with a similar instrumentation recently described by Magnani, Centritto & Grace 1996) that allows field measurements of KR in plants with trunks up to 60–80 mm in diameter.

The root hydraulic conductance of three to five different plants growing in the field was measured in April, July and October 1997, using a HPFM. The instrument and its theory of operation was first described by Tyree et al. (1995). Briefly, the HPFM consists of an apparatus that allows the perfusion of water into the base of a root system or a shoot while rapidly changing the applied pressure (P) and simultaneously measuring the corresponding flow (F).

Shoots were excised from the roots (about 150 mm above the soil) under water by constructing a watertight container around the base of each plant. The HPFM was connected to the root system and three to six transient flow measurements were immediately made. During each measurement, P was increased at a rate of 5–8 kPa s–1 while measuring the instantaneous flow every 3 s. The entire sequence of flow measurements took about 5 min.

The flow was then plotted versus the applied pressure driving the flow into the root system. The curves were nonlinear at the first 0·2–0·3 MPa pressure applied but then became distinctly linear. The slope of this linear region was found to decrease in the first two to three transient measurements and became constant during the subsequent two to three measurements. Root hydraulic conductance (KR) was calculated from the slope of the linear region of the last transient measurement, by linear regression of the data (Tyree et al. 1995).

All the leaves from the excised shoots were enclosed in plastic bags, put in a thermal bag at about 4 °C and transported to the laboratory where total leaf surface area (AL) was measured using a leaf area meter (LI-3000 A; LiCor Inc.) equipped with a belt conveyer (LI-3050 A). The parameter KR was normalized by AL, thus obtaining the root hydraulic conductance per unit leaf surface area (KRL, kg s–1 m–2 MPa–1). Such a procedure had the aim of providing an estimate of the ‘sufficiency’ of the root system to supply leaves with water (Tyree, Velez & Dalling 1998). We are aware that KR should be normalized to the root surface area (AR) for a more correct evaluation of KR in physical terms. Nonetheless, when working in the field with plants of some years of age, the intrinsic error in measuring the water absorbing AR of excavated root systems may be very large.

Estimating the water balance of plants

The water balance of plants is controlled by the relative rates of water absorption and water loss under a given water potential gradient along the soil–plant–atmosphere continuum. In an attempt to compare the water balances of Q. suber and Q. cerris plants under drought conditions, the ratio of gL to KRL was calculated, with gL being measured in mmol s–1 m–2 and then converted into kg s–1 m–2. As KRL was measured in kg s–1 m–2 MPa–1, the ratio gL/KRL was in units of MPa.

The gL/KRL ratio (‘conductance ratio’, CR) gives the change in water potential across the root system needed to extract water from the soil to keep up with a unit of evaporative driving force. This is because the water vapour flux E = gLΔW, where ΔW is the water vapour mole fraction difference between the evaporating leaf surface and the bulk air. The water flow through the root system (normalized to leaf area) is E = KRLΔΨ where ΔΨ is the water potential difference between the soil and the shoot base where the roots end. At steady flow across the plant, gLΔW = KRLΔΨ, and hence gL/KRL = ΔΨ/ΔW. A plant suffering drought, might drop KRL more than gL, and this would result in a high CR.

The theory of the HPFM technique for measuring KR (Tyree et al. 1995) implies that the relation between the flow through the roots to the pressure driving the flow becomes linear once all the elastic components of the instrument–root system have been eliminated, including any eventual xylem emboli. Under these conditions, the HPFM should measure the maximum KRL for a given root system at a given time. Taking this into account, the gL used for calculating CR was the mean of the maximum diurnal values recorded in a given month.

RESULTS

Seasonal changes in the soil moisture volumetric content measured at various depths in the experimental site are reported in Table 2. As expected, the highest soil water content was recorded in April, ranging between 12 and 16%. In summer, it dropped to 5–7% along the upper 0·6 m explored. A vertical gradient of soil moisture content was recorded in October when the soil moisture content was about the same as in April at 0·15 and 0·3 m depth but it was only 8–10% at 0·45 and 0·6 m depth. This was the result of the early autumn rainfall which was typically brief, heavy showers that wetted the upper soil layers but did not penetrate deep into the soil.

Table 2.  . Volumetric soil moisture content (%) measured at different soil depths. Means are given ± 1 SD (n = 10) Thumbnail image of

Leaf water status

The diurnal time courses of gL, ΨL and RWC of Q. suber and Q. cerris leaves measured in April (open circles), July (solid squares) and October (open triangles) are reported in Figs 1 and 2, respectively.

Figure 1.

. Diurnal time course of leaf conductance to water vapour (gL), leaf water potential (ΨL) and relative water content (RWC), measured in April (open circles), July (solid squares) and October (open triangles). Vertical bars indicate 2 SD.

Figure 2.

. Diurnal time course of leaf conductance to water vapour (gL), leaf water potential (ΨL) and relative water content (RWC), measured in April (open circles), July (solid squares) and October (open triangles). Vertical bars indicate 2 SD.

In the spring, the gL measured in Q. suber (Fig. 1) was, on average, 140 mmol s–1 m–2 between 0730 h and 1830 h, with three distinct peaks up to 220 mmol s–1 m–2 recorded at 0930, 1430 and 1630 h. Such peaks of gL produced opposite changes in ΨL and RWC (Fig. 1). With sufficient soil water availability (April), the diurnal ΨL was high, i.e. between –0·45 and –0·9 MPa and the leaf RWC was as high as 86–92%.

In summer, the gL of Q. suber leaves dropped significantly to average diurnal levels of about 90 mmol s–1 m–2 with highest values of 115 mmol s–1 m–2 recorded during the warmest hours of the day. In spite of this lower gL, leaf RWC decreased to a nearly constant level of about 82% and ΨL was reduced to –1·7 MPa. It should be noted, however, that the leaves did not reach the turgor loss point. In fact, ΨL at the turgor loss point, as derived from the PV curves measured in July (Fig. 3), was much more negative (ΨTLP was –2·6 MPa in July, Table 3) and the minimum diurnal ΨL of –1·7 MPa corresponded to a turgor pressure (PT) of 0·5 MPa, indicating a PT drop of about 75%.

Figure 3.

. Typical Höfler diagrams showing changes in leaf turgor pressure (PT), water potential (–ΨL) and osmotic potential (–π) as a function of symplasmic water loss, measured in July 1997 in 1-year-old leaves of Quercus suber and Quercus cerris.

Table 3.  . Leaf osmotic potential at full turgor (ΨΠo), leaf water potential at the turgor loss point (ΨTLP) and bulk modulus of elasticity at full turgor (ɛmax). Means are given ± 1SD (n = 5) Thumbnail image of

Even lower gL values were recorded in October (Fig. 1) when the mean diurnal gL was found to be only about 50 mmol s–1 m–2, corresponding to a ΨL value as high as –0·5 MPa and a RWC value of about 88%. In other words, the major impact of the Mediterranean dry summer on the hydration of Q. suber leaves was leaf RWC values that were only 8% less than those recorded in the spring.

The seasonal dynamics of leaf water status recorded in Q. cerris (Fig. 2) were quite different from those of Q. suber. In April, the mean diurnal gL was much lower in Q. cerris than in Q. suber (less than a half) while in summer and autumn the two oak species showed values of gL in very close agreement.

The diurnal RWC of Q. cerris leaves was as high as 96% in the spring and 90% in autumn but in the summer it decreased to 74%. Accordingly, the time course of ΨL as recorded in July showed a substantially reduced diurnal value of –2·4 MPa compared with a minimum diurnal ΨL of –0·75 MPa measured in April and October. Although the leaves of Q. cerris did not reach the turgor loss point, even in summer (ΨTLP as derived from the PV curves was –3·1 MPa, Table 3), they retained much less of their turgor than leaves of Q. suber, under the same environmental conditions (Fig. 3, Table 3). In fact, the minimum diurnal ΨL measured in July (–2·4 MPa) would correspond to a PT of 0·25 MPa, i.e. the leaves would have undergone a PT drop of over 90%.

Data resulting from the PV curves (Table 3) also showed that the leaves of Q. suber increased their solute content slightly through the year (Ψπο was reduced from –1·82 MPa in April to –2·28 MPa in October, i.e. a decrease of about 20%). The osmotic adjustment of Q. cerris leaves was much larger with Ψπο reduced from –1·14 MPa in April to –2·33 MPa in July (Table 3). The slight but significant increase in Ψπο of Q. cerris leaves in October (–1·98 MPa) with respect to that recorded in July, was probably the result of the beginning of solute export from the leaves before shedding.

The bulk modulus of elasticity at full turgor (ɛmax) (Table 3) was about twice as high in Q. suber as in Q. cerris. Typical PV curves measured in Q. suber and Q. cerris leaves in July (Fig. 3) showed the influence of a higher ɛmax on ΨL changes during leaf dehydration. As an example, a water loss of only 5% caused ΨL to drop to –1·0 MPa in Q. suber but to only –0·5 MPa in Q. cerris. Accordingly, the relative symplasmic water loss at PT = 0 was about 18% in Q. suber but as high as 27% in Q. cerris leaves.

Root hydraulic conductance

Seasonal changes in KRL (Fig. 4) showed opposite seasonal time courses in the two oak species.

Figure 4.

. Root hydraulic conductance normalized by the leaf surface area (KRL) measured in April (white columns), July (black columns) and October (dashed columns) 1997. Vertical bars indicate 1 SD.

The root systems of Q. suber had values of KRL between 4·8 (white columns) and 6·5 x 10–4 kg s–1 m–2 MPa–1 (black columns) in April and July, respectively, with no significant differences between the two months. In October, KRL increased to 13·2 x 10–4 kg s–1 m–2 MPa–1, i.e. a nearly twofold increase (dashed columns, Fig. 4).

By contrast, the highest annual KRL of Q. cerris plants was recorded in April (52·4 x 10–4 kg s–1 m–2 MPa–1). Then, KRL dropped to about 2·0 x 10–4 kg s–1 m–2 MPa–1 between July and October, i.e. the drop in KRL was over 90%. In other words, when KRL was highest in Q. suber it was lowest in Q. cerris and vice versa.

It is worth noting that during the dry Mediterranean summer, the root systems of Q. suber maintained KRL values over three times higher than those recorded in Q. cerris (6·5 x 10–4 versus 1·8 x 10–4 kg s–1 m–2 MPa–1).

Conductance ratio

The gL/KRL ratio (= CR) showed opposite seasonal changes in the two oaks. In particular, CR as calculated for Q. suber, was highest in April (8·22, white columns, Fig. 5) and decreased to 3·27 in July and further to 0·98 in October, because of lower gLs recorded in July and even lower gLs and the highest KRLs in October.

Figure 5.

. Ratio of leaf conductance to water vapour (gL) to root hydraulic conductance scaled to leaf surface area (KRL) calculated for Quercus suber and Quercus cerris plants on the basis of gL and KRL (see text) measured in April (white columns), July (black columns) and October (dashed columns). Vertical bars indicate 1 SD.

It is of interest to note that Q. suber was growing most actively in October with most new leaves being produced then and visible twig elongation. Unfortunately, it was not possible to include growth measurements in the present study.

By contrast, CR as calculated for Q. cerris was lowest in the spring (0·31), increased to 12·42 in July and decreased to 7·37 in October. Apparently, it was the dramatic drop in KRL measured in July and October that determined the observed high values of CR in these months.

DISCUSSION

The strategies for drought resistance adopted by the evergreen, sclerophyllous Q. suber and by the deciduous, soft-leaved Q. cerris were quite different although the trees had apparently been growing at the same site for a long time.

The consistent decrease in gL measured in the Q. suber trees during the dry period (July) suggested an efficient stomatal control of transpiration by this species. At the same time, the high ɛmax of the leaves, i.e. the high ‘rigidity’ of the living cell walls (Table 3), allowed ΨL to drop rapidly and substantially as soon as the leaves began to lose water in the early hours of the day (Fig. 1). If changes in ΨL resulting from the PV curves are compared with the field-recorded ΨL changes (Fig. 1), it can be calculated that a ΨL drop from –0·6 to –1·4 MPa, like that recorded between 0830 and 1130 h, corresponds to a relative symplasmic water loss as small as 3–7% (Fig. 3).

It has been suggested that the capability of dropping ΨL rapidly and transiently in response to small water losses is of advantage to plants subjected to short-term (e.g. diurnal) water stress in that they can thereby enhance the driving force for water extraction from the soil, thus maintaining leaf hydration homeostatically high (Kikuta et al. 1997; Lo Gullo & Salleo 1988). Of course, this is only possible if roots maintain a sufficient hydraulic conductance. If, in contrast, the roots develop a high resistance to flow with increasing drought, the water potential drop throughout the plant xylem would increase beyond the cavitation threshold, thus leading to diffuse xylem embolism.

In summer, Q. suber plants maintained values of KRL at the same level as in the spring when the soil moisture content was three times higher. The CR calculated for Q. suber plants was low in summer. This suggests that this species maintained a favourable ratio between water loss and uptake during the dry period. Such behaviour is quite similar to that described for Laurus nobilis L. (a Mediterranean evergreen sclerophyll, Lo Gullo & Salleo 1988) which has been classified as a case of ‘drought avoidance’ (Levitt 1980).

Quercus cerris plants behaved quite differently under water shortage conditions. Although the gL values in July were similar to those measured in Q. suber, they were significantly higher than in April, thus indicating that little or no stomatal control of transpiration was operating in the dry period. Moreover, the quite low values of KRL measured in July (Fig. 4) contributed to cause large leaf dehydration (the value of RWC dropped to 74% and ΨL at 0730 h was about –2·05 MPa). The higher soil moisture content measured in autumn (at least in the upper soil layers) provided sufficient water to the leaves to allow the RWC to rise to 90%, even in the presence of root systems with persistent low hydraulic conductances. Because no visible damage was observed in summer to leaves of the young plants of Q. cerris, although they had lost over 90% of their turgor pressure on a diurnal basis, the species was considered to be ‘drought tolerant’ at least within the limits of drought occurring in the area studied.

An interesting aspect of the changes in the leaf water status of the two species was the discrepancy observed in Q. suber, between the symplasmic water loss (SWL) (Fig. 3) as compared to the leaf relative water deficit (RWD, i.e. 100 –RWC, Fig. 1) at equal ΨL values. Discrepancies between SWL and RWD were really strong in Q. suber leaves in that a ΨL value of –1·7 MPa corresponded to a SWL of 8% and a RWD as high as 19%. Similar discrepancies occurred for most of the measured ΨL values in this species whereas there was quite good agreement between SWL and RWD values in Q. cerris. A previous study (Kikuta et al. 1997) showed that during leaf dehydration (or, during leaf pressurization in the pressure chamber) the leaf's apoplast (consisting of mechanical tissue and xylem conduits) underwent extensive cavitation. The fact that the values of RWD measured in the field were constantly higher than the SWL values derived from the PV curves at equal ΨL values in Q. suber, suggests that the major part of water loss was from the leaf apoplast as a consequence of cavitation. This, has been suggested to play a role in buffering the leaf water content and would attribute an adaptive significance to the sclerophyllous habit of Mediterranean evergreens (Salleo, Nardini & Lo Gullo 1997). In contrast, the values of SWL and RWD coincided fairly well in the soft leaves of Q. cerris. This is in accordance with the basic theory of the pressure chamber as was advanced for soft-leaved species (Tyree & Hammel 1972).

The water balance of Q. cerris was most favourable in the spring (CR was less than unity) coinciding with bud-break and shoot sprout but it became strongly imbalanced during the summer (CR was 12·42) when water loss was not balanced by uptake. The persistently high CR values calculated in autumn were the result of the apparent lack of any late summer growth in Q. cerris.

The highest CR calculated for Q. suber in the spring was not the expression of water stress conditions because the soil moisture content was high enough to allow adequate water supply to the leaves while plants showed some production of new leaves so that only some additional water demand was necessary for tissue expansion.

The most interesting results from calculating conductance ratios can be summarized as follows: firstly, a comparison of the seasonal time course of KRL with that of CR strongly suggests that CR was mainly influenced by KRL and less by gL values. This suggests that the resistance to environmental stresses in plants cannot be correctly evaluated in the absence of reliable measurements of the hydraulic behaviour of roots. In this sense, the HPFM was shown to be a very useful instrument in that it enabled numerous reproducible measurements to be made in a matter of minutes. Nonetheless, further development of HPFM is necessary to overcome some of the major intrinsic operational limits of this instrument (Tyree et al. 1995). Secondly, both Quercus species showed the lowest CR values while growing most actively, i.e. in October for Q. suber and in April for Q. cerris; and thirdly, during the dry summer period, the CR in Q. suber was only about 25% of that of Q. cerris (about three versus 12, Fig. 5). In accordance with the field-recorded parameters describing leaf water status, the lower CR calculated for Q. suber confirms that this species was more competitive than Q. cerris during water shortage. This is a reasonable conclusion because Q. cerris usually grows in more humid areas and on deeper soils than Q. suber (Pignatti 1982). Nonetheless, no real competition for available water existed between the two oaks, one of which (Q. suber) continued to extract sufficient water from the soil to maintain leaf hydration while the other was unable to do the same and suffered dehydration.

As was previously observed in the case of the drought-tolerant wild olive tree and of the drought-avoiding carob tree (Lo Gullo & Salleo 1988; Salleo & Lo Gullo 1993), the drought-avoiding Q. suber and the drought-tolerant Q. cerris were shown to be adapted to cohabit the same dry site. We do not know whether the presence in the same area of species with these two strategies for drought resistance can be generalized but this seems to be a promising starting point for further studies of physiological adaptation within ‘plant communities’.

ACKNOWLEDGEMENTS

We wish to thank the Forest Service of the Sicily Region, and in particular the personnel of the Forestry Station of Caronia for assistance. We are also grateful to Professor M. T. Tyree for stimulating criticism and discussion and to the anonymous reviewers for valuable suggestions and encouragement. This paper was supported by a grant from the Italian Ministry for University and Scientific and Technological Research (National Projects).

Ancillary