Tree Physiology 23, 747–758 © 2003 Heron Publishing—Victoria, Canada Sap flow of three co-occurring Mediterranean woody species under varying atmospheric and soil water conditions JORDI MARTÍNEZ-VILALTA,1–3 MARTA MANGIRÓN,4 ROMÀ OGAYA,4 MIQUEL SAURET,2 LYDIA SERRANO,4 JOSEP PEÑUELAS 4 and JOSEP PIÑOL2 1 Present address: Ecology and Resource Management, School of Earth, Environmental & Geographical Sciences, University of Edinburgh, Darwin Building, King’s Buildings, Edinburgh EH9 3JU, U.K. 2 CREAF, Facultat de Ciències, Universitat Autònoma de Barcelona, Bellaterra-08193 (Barcelona), Spain 3 Author to whom correspondence should be addressed (Jordi.Martinez-Vilalta@ed.ac.uk) 4 Unitat Ecofisiologia, CSIC-CREAF, CREAF, Facultat de Ciències, Universitat Autònoma de Barcelona, Bellaterra-08193 (Barcelona), Spain Received December 19, 2001; accepted January 24, 2003; published online July 1, 2003 Keywords: Arbutus unedo, drought, Phillyrea latifolia, Quercus ilex, sap flux, water use. Introduction The climate in the Mediterranean basin is characterized by an acute summer drought that, according to most climatic scenarios, will intensify as a result of climate change (IPCC 2001). In this context, a proper understanding of the effects of water shortage is required if predictions of the impact of climate change on Mediterranean vegetation are to be made (Borghetti et al. 1998). The long-term measurement of water use of cooccurring species in relation to environmental variables is a powerful approach for studying the response of plants to water availability, particularly when the species studied show contrasting adaptations to drought (e.g., Pataki et al. 2000). Because plant responses are likely to be influenced by different environmental variables depending on the temporal scale considered (Pataki et al. 1998), it is important that data are gathered at a sufficiently high temporal resolution. This requirement is fulfilled by sap flow studies because they provide estimates of transpiration with a temporal resolution of less than one day (Granier et al. 1996). Evergreen forests dominated by holm oak (Quercus ilex L.) are one of the most important vegetation types in the Mediterranean basin. In Spain, for example, holm oak forests constitute 25% of the total forested area (Terradas 1999). Arbutus unedo L. and Phillyrea latifolia L. co-occur in holm oak forests throughout the Mediterranean, typically forming a lower tree layer under the Q. ilex canopy. Although the three species are evergreen and have similar distributions (Bolós and Vigo 1990–1995), previous studies have shown that they differ markedly in water relations and resistance to drought (Table 1). In contrast to the other two species, P. latifolia exhibits little stomatal control over water loss and maintains physiological activity even when leaf water potentials are low (Tretiach 1993, Castell et al. 1994, de Lillis and Mirgone 1994, Peñuelas et al. 1998). Within a site, P. latifolia tends to occupy the driest areas, whereas A. unedo is usually restricted to the most humid microenvironments. However, recent Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Summary We studied the seasonal patterns of water use in three woody species co-occurring in a holm oak forest in northeastern Spain. The three species studied, Quercus ilex L., Phillyrea latifolia L. and Arbutus unedo L., constitute more than 99% of the total basal area of the forest. The study period included the dry seasons of 1999 and 2000. Water use was estimated with Granier-type sap flux sensors. Standard meteorological variables, soil water content and leaf water potentials were also monitored. All monitored individuals reduced leafrelated sap flow (Ql) during the summer, concurrent with an increase in soil moisture deficit (SMD). Despite similar maximum Ql between species, the decline in Ql with increasing SMD was species-dependent. The average reduction in Ql between early summer and the peak of the drought was 74% for A. unedo (n = 3), 58% for P. latifolia (n = 3) and 87% for Q. ilex (n = 1). The relationship between canopy stomatal conductance (Gs) and vapor pressure deficit (D) changed during the course of the drought, with progressively lower Gs for any given D. Summertime reductions of Ql and Gs were associated with between-species differences in vulnerability to xylem embolism, and with the corresponding degree of native embolism (lowest in P. latifolia and highest in Q. ilex). Our results, combined with previous studies in the same area, outlined differences among the species studied in manner of responding to water shortage, with P. latifolia able to maintain water transport at much lower water potentials than the other two species. In an accompanying experiment, A. unedo responded to an experimental reduction in water availability by reducing Ql during the summer. This species also modified its water use between years according to the different seasonal patterns of precipitation. These results are discussed in relation to the possible impacts that climate change will have on Q. ilex-dominated forests. 748 MARTÍNEZ-VILALTA ET AL. Table 1. Mean (± 1 SE) xylem hydraulic and water-use-related parameters for Arbutus unedo, Phillyrea latifolia and Quercus ilex at the study site. Different letters indicate significant differences between species. Parameter Tissue Leaf:sapwood area ratio (× 10 –2 m2 m –2) 1 Carbon isotope discrimination (δ13C; ‰)2 Specific conductivity (× 10 4 m2 MPa –1 s –1) 3 Leaf-specific conductivity (× 10 7 m2 MPa Xylem pressure at 50% embolism (MPa) 3 –1 Root Stem s –1) 3 Root Stem Arbutus unedo Phillyrea latifolia Quercus ilex 9.5 ± 0.5 –25.0 ± 0.3 15.9 ± 5.1 5.4 ± 1.1 2.3 ± 0.5 –1.2 ± 0.2 –3.1 ± 0.5 8.8 ± 0.3 a –27.0 ± 0.4 b 18.9 ± 8.9 a 4.1 ± 0.8 a 2.2 ± 0.5 a –5.3 ± 0.9 b –6.6 ± 1.3 b 24.2 ± 4.2 b –25.9 ± 0.3 b 60.1 ± 37.0 a 25.0 ± 2.2 b 12.0 ± 1.7 b –1.2 ± 0.5 a –2.0 ± 0.4 a a a a a a a a 1 Data from this study and from Ogaya et al. (2003). M. Mangiron (unpublished results). 3 Martínez-Vilalta et al. (2002b). 2 Materials and methods Study site The site was located in a forested area in the Prades Mountains, NE Spain (41°13′ N, 0°55′ E; 990 m a.s.l.). The climate is Mediterranean, with a mean annual rainfall of 537 mm (1981–1995) and moderate temperatures (10.0 °C mean at Prades, 1000 m a.s.l.). Additional information about the study area can be found in Hereter and Sánchez (1999). Experimental plots were located on the south-facing upper slopes of the Torners valley (about 35% slope). The substrate is fractured schist, and soils are xerochrepts with a clay-loam texture. Soil depth is 65 ± 3 cm. The forest is dominated by Q. ilex, P. latifolia and A. unedo, which constitute ≥ 99% of the total basal area in the two study plots (Table 2). Canopy height is 4–7 m and LAI is ≈ 4.5 (Sala et al. 1994). This community is known to be limited by water availability (Rodà et al. 1999). The study period included the dry seasons of years 1999 and 2000 (Figure 1). Experimental design This study was part of a drought simulation experiment, which included eight 15 × 10 m plots (four control and four droughtexposed), located at the same altitude along the slope. However, sap flow sensors were installed in only two of the plots (one assigned to each treatment), and we focused only on those in this study. The distance between the edges of these two plots was about 7 m. In the drought treatment, water availability to trees was restricted from March 1999 until the end of the study period by plastic sheets and collectors suspended 50–100 cm above the soil, and 1-m deep trenches on the upper sides of the plots to exclude lateral water flow. The plastic structures were made with transparent PVC and covered 30% of the surface area of the plots. The effectiveness of plastic sheets in excluding throughfall was tested in the drought-exposed plot by monitoring soil water content with three groups of paired time-domain-reflectometry (TDR) sensors over the entire study period. Within each group, one sensor was placed under a plastic sheet and the other in an adjacent, uncovered position. Water content was, on average, 3% units lower under the sheets than in the uncovered positions. Although the effect was not significant when integrated over the study period (P = 0.15, two-way ANOVA with position and sampling date as repeated measurement factors), it became significant after the rain events (P < 0.05, paired t-tests). TREE PHYSIOLOGY VOLUME 23, 2003 Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 droughts in eastern Spain, particularly in summer 1994, have shown that Q. ilex is closer to its limit in coping with water stress than the other two species. In some areas, 80% of Q. ilex individuals lost all their foliage as a result of the 1994 drought (Lloret and Siscart 1995). Most of these trees resprouted, although mortality was about 15% in many populations (Peñuelas et al. 1998, 2000). Therefore, it has been hypothesized that Q. ilex may be substituted for P. latifolia in the most water-limited areas of the Mediterranean if the climate becomes drier (Peñuelas et al. 1998, 2000, Martínez-Vilalta et al. 2002a). A recent study (Martínez-Vilalta et al. 2002b) has shown that A. unedo, P. latifolia and Q. ilex have contrasting hydraulic properties, which partially explain the observed differences in water relations and drought resistance. In particular, P. latifolia is much more resistant to xylem embolism than the other species, whereas A. unedo is slightly more resistant than Q. ilex (Table 1). In the present study, we used daily sap flow measurements to evaluate seasonal patterns of water use in these species, with the aim of integrating previous knowledge on their water relations and understanding the underlying mechanisms that explain their contrasting responses to seasonal drought. It was hypothesized that: (1) consistent with the previous considerations, the reduction of sap flow during a summer drought would follow a pattern of Q. ilex > A. unedo > P. latifolia; (2) the relationship between canopy stomatal conductance and vapor pressure deficit (D) would change accordingly during the drought period, with Q. ilex and A. unedo showing the largest increase in stomatal sensitivity to D; and (3) because water is a limiting resource in the community studied, its use would respond to differences in seasonal water availability between years and to an experimental reduction of water availability. SAP FLOW OF THREE MEDITERRANEAN WOODY SPECIES 749 Table 2. Tree density (stems ha –1) and basal area (m 2 ha –1) at the two plots studied. Only adult individuals with a diameter > 2 cm (measured at 0.5 m height) were counted. Species Quercus ilex L. Phillyrea latifolia L. Arbutus unedo L. Calicotome espinosa (L.) Link Juniperus oxycedrus L. Rhamnus alaternus L. Sorbus torminalis (L.) Crantz Sorbus domestica L. Total Control plot Treatment plot Density (stems ha –1) Basal area (m2 ha –1) Density (stems ha –1) Basal area (m2 ha –1) 9467 3667 2200 133 133 67 0 0 15667 33.87 3.46 5.40 0.06 0.06 0.03 0 0 42.88 7733 3533 2133 0 0 0 200 133 13733 32.95 3.13 7.90 0 0 0 0.31 0.15 44.44 Atmospheric and soil moisture measurements Sensors for measuring temperature and relative humidity (Model 50Y, Campbell Scientific, Logan, UT), solar radiation (Campbell SP1110 pyranometer) and wind speed (Campbell A100R switching anemometer) were located on a mast at approximately crown height. The mast and a standard rain gauge (Campbell ARG100 tipping bucket rain gauge; 1.5 m height) were situated in a clearing within 40 m of the measured trees. Data, stored on a data logger (Campbell CR10X), were sampled at 5-s intervals and averaged every 30 min. Vapor pressure deficit was calculated every 30 min from relative humidity and temperature. Data were converted to daily means. Hourly evapotranspiration (ET) was calculated with the Penman-Monteith equation, assuming a constant canopy resistance of 70 s m –1 (Allen et al. 1998). Soil moisture deficit (SMD) was calculated as the difference between cumulative ET and cumulative rainfall, with SMD < 0 set to zero. It was assumed that on January 1, 1999, after 77 mm of precipitation during the previous week, SMD = 0. Soil moisture deficit was calculated from ET rather than from actual estimates of canopy transpiration because sap flow measurements made during the 2-year study were too fragmentary. Soil water was monitored by TDR every 2–3 weeks throughout the study. Three-rodded steel sensors were located randomly throughout the study plots, and volumetric water content at 0–25 cm (θ0–25; 8 sensors per plot) and at 0–40 cm (θ0–40; two sensors per plot) was estimated with a cable tester (1502B, Tektronix, Beaverton, OR). Leaf water potential Leaf water potentials were measured seasonally throughout the study period with a pressure chamber (PMS Instruments, Corvallis, OR) (Scholander et al. 1965). On each sampling date, shoot tips from the trees monitored with sap flow sensors or adjacent individuals (n ≥ 3 individuals per species, one shoot tip per individual) were measured at predawn ( just before sunrise) (0300–0500 h, solar time) and at midday (1100–1300 h). The time lag between shoot excision and measurement was always < 1 min. Shoot tips were not bagged before measurement. Arbutus unedo individuals were measured only during the year 2000. Sap flow Stem flux was monitored with 20-mm-long, constant heat flow gauges constructed according to Granier (Granier 1985, 1987). Within each of the two study plots, sap flow of three dominant trees per species was monitored (Table 3). Sensors were installed between November 1998 and February 1999, except those on A. unedo individuals in the drought-exposed plot, which were installed in October 1999. A probe pair was inserted radially into the stem of each tree at breast height after removing the bark to expose the outer surface of the sapwood. The vertical separation between probes was approximately 15 cm. The temperature difference between the probes was recorded continuously to obtain sap flux density by means of the equation derived empirically by Granier (1985, 1987). Daily maximum temperature differences were determined from stored data and used to estimate temperature differences under zero flow conditions. This variable was nearly constant during the study (the slope of the relationship between the daily maximum electromotive force produced by the thermocouples and time was between –1 × 10 – 4 and +1 × 10 – 4 mV day –1; the coefficient of variation of maximum electromotive force was 3.38 ± 0.49%), suggesting that no substantial variation in the thermal properties of the surrounding wood occurred during the measurements. Sensors and trunks were insulated with glass wool to minimize ambient temperature gradients, and sensors were oriented to the north to avoid azimuthal effects TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 The composition of the plant community was similar at the two plots studied (Table 2). Photosynthetically active radiation (PAR) intercepted by the canopy was measured with a ceptometer (Decagon Sunfleck Ceptometer, Pullman, WA) in each plot. The results indicated no between-plots difference in either initial leaf area index (LAI) or in LAI dynamics during the study (data not shown). Seasonal variation in intercepted PAR was low, with a maximum variation of 15% measured over the study period in one of the plots and 10% in the other, suggesting that LAI was approximately constant. 750 MARTÍNEZ-VILALTA ET AL. rected according to Clearwater et al. (1999). Because the species studied are diffuse-porous, when sapwood depth exceeded sensor length, sap flux density was assumed to be uniform over the entire sapwood depth. However, this assumption is not critical because we were interested mainly in the comparison of sap flow among species and, in most cases, in the comparison of values from the same trees at different periods. The area of active sapwood was obtained by averaging the thickness of sapwood at both ends of the core. The allometric relationships obtained in the same study area by Ogaya et al. (2003) for P. latifolia and Q. ilex, and by Lledó (1990) for A. unedo, were used to calculate leaf-related sap flow (Ql; kg m –2 day –1). Diurnal values of Ql were summed to daily values to avoid the complications introduced by stem capacitance (Oren and Pataki 2001). Nomenclature follows that of Edwards et al. (1996). Whole-plant hydraulic conductance and canopy stomatal conductance kS-L = Ql,max/(ΨPD – ΨMD) (1) where ΨPD is predawn water potential, ΨMD is midday water potential and Ql,max is leaf-related sap flow during peak transpiration at midday. Canopy stomatal conductance (Gs; mm s –1) can be derived from sap flow measurements, based on the assumption that sap flux density scaled by sapwood-to-leaf area ratio (i.e., Ql) is equal to transpiration per unit of leaf area. For species with small leaves, with strong aerodynamic coupling to the atmosphere, the Penman-Monteith equation can be simplified, leading to the following relationship (Whitehead and Jarvis 1981, Pataki et al. 1998): Gs = Figure 1. Daily values of maximum and minimum air temperature (Tair), cumulative water balance (R = rainfall; ET = potential evapotranspiration; and SMD = soil moisture deficit) and volumetric soil water content (θ) at two different depths during 1999 and 2000 at the study site. Shaded areas indicate the periods in which sap flow was monitored. Bars for θ represent standard errors. (Oliveras and Llorens 2001). Sap flow data were sampled at 5-s intervals and recorded by a data logger. The thickness of active sapwood was determined with an increment corer at the end of the study. The depth of active sapwood was estimated from the translucency of the wood in A. unedo and P. latifolia, and by stereoscopic examination in Q. ilex. Sapwood depth ranged between 18.7 and 35.8 mm (Table 3). In the only case where sapwood depth was less than the 20-mm sensor length, the temperature difference was cor- γ λQl ρc p D (2) where γ is the psychrometric constant (kPa K –1), λ is the latent heat of vaporization of water (J kg –1), ρ is the density of air (kg m –3) and cp is the specific heat of air at constant pressure (J kg –1 K –1). We used daily means of all quantities, excluding values of D < 0.1 kPa (Phillips and Oren 1998). In our case, the assumption of strong coupling to the atmosphere was tested by comparing Gs calculated from Equation 2 with boundary layer conductance (gbl), estimated from daily mean wind speed (u; m s –1) using gbl = 6.62(u/d )0.5 (Jones 1992). Based on the maximum leaf dimension (d) of 0.01 m, the ratio gbl /Gs was > 10 for more than 95% of days in all studied trees, supporting the conclusion that calculated Gs was dominated by stomatal aperature rather than by the boundary layer (Whitehead and Jarvis 1981). Leaf size and shape were similar in the species TREE PHYSIOLOGY VOLUME 23, 2003 Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 When data were available, measurements of leaf water potentials were combined with Ql to calculate whole-plant hydraulic conductance (kS-L; kg m –2 MPa –1 s –1; Wullschleger et al. 1998): SAP FLOW OF THREE MEDITERRANEAN WOODY SPECIES 751 Table 3. Main structural characteristics of trees monitored with sap flow sensors. Only those trees considered in the analysis (data available for > 80% of days during the study period) are included in the table (see text). Abbreviations: DBH = diameter at breast height; and sw = sapwood. Tree ID Species Treatment Operating period DBH (cm) Leaf area (m2) Depth sw (mm) 17 21 16 12 10 7 14 18 19 Arbutus unedo Arbutus unedo Arbutus unedo Arbutus unedo Arbutus unedo Phillyrea latifolia Phillyrea latifolia Phillyrea latifolia Quercus ilex Control Control Control Drought Drought Drought Control Control Control 1999 and 2000 1999 and 2000 1999 and 2000 2000 2000 1999 1999 1999 1999 6.3 7.4 8.4 8.4 9.5 4.1 6.5 5.9 9.6 2.23 3.31 4.20 4.23 6.14 0.94 2.26 1.90 7.53 23.4 34.8 35.0 24.3 32.0 18.7 29.5 26.4 35.8 studied. In applying Equation 2, we assumed that there was no vertical gradient of D throughout the canopy. Data analysis Ql (%) = 100e a (SMD – SMD 0) b (3) where SMD0 is the soil moisture deficit at the beginning of the summer period, and a and b, the fitted parameters, measure the sensitivity of Ql to SMD. When studying the relationship between Ql and SMD, summer days with mean daytime D < 1 kPa were removed from the analysis to avoid uninformative dispersion. Differences between curves were tested among species, treatments, or years by means of F-tests comparing the mean squares of the curvilinear regression with and without segregating data according to the variable under consideration (Potvin et al. 1990). When comparing species, periods or years, only individuals from the control plot were used, but whenever the treatment effect was clearly nonsignificant (P > 0.15), data were pooled by species. Comparisons were made only if there were at least two replicates within each level of the factor under consideration. Estimates of Gs were regressed against D by means of a logarithmic function (Schäfer et al. Results Seasonal water relations and treatment effects The dynamics of soil water content (θ) showed the typical pattern, with lowest values during the summer months of both years coinciding with periods with highest temperature and potential evapotranspiration (Figure 1). Water content was, on average, 9% units higher in the first 25 cm of soil than in the uppermost 40 cm (Figure 1c), probably because of higher porosity in the upper soil layers (J. Piñol, unpublished results). Considering all experimental plots, there was a significant reduction in θ0–25 associated with the drought treatment (P = 0.03; three-way, nested ANOVA), but no difference was found between the plots with sap flow sensors (P = 0.28). This was because the soil of the drought-treated plot was originally wetter than that of the control. However, the treatment was associated with a significant change in the difference in θ0–25 between the plots (P < 0.01; t-test comparing the difference before and after the onset of the treatment), causing the drought-exposed plot to be consistently drier than the control (Figure 2, maximum difference of about 15%). There was no significant difference in water potentials between treatments for P. latifolia or Q. ilex (P > 0.17 for ΨPD and P > 0.13 for ΨMD), and thus the values were pooled by species in the subsequent analysis. In contrast, A. unedo ΨPD was significantly lower in the dry plot than in the control plot (P = 0.02) during spring–summer of 2000 (the only period in which this species was measured). No treatment effect was observed in ΨMD or in the difference between ΨMD and ΨPD for A. unedo. Among species, ΨPD was significantly lower in Q. ilex than in A. unedo (P < 0.01), and lower in P. latifolia than in Q. ilex TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Soil water content, leaf water potentials and kS-L were compared among treatments or among species using repeated measurements ANOVA. Because we were interested in the drought response of the studied species, our analysis of sap flow was restricted to measurements conducted from late spring (period with maximum flow) to late summer/early autumn (period with the lowest values). Because of sensor malfunction, the analysis of sap flow was limited to control plots in 1999 (n = 3 individuals for A. unedo and P. latifolia and n = 1 for Q. ilex) and to A. unedo in 2000 (n = 3 individuals in the control plot and n = 2 in the drought-exposed one) (Table 3). Only trees in which sap flow data were available for > 80% of days during the study were considered. Leaf-related sap flow and Gs were averaged by species. Summer reductions of Ql were compared between species by fitting the relationship between Ql (%) (in relation to the species maximum reached at the beginning of the summer period) and SMD with the following function: 2000, Oren and Pataki 2001). Although this procedure suffers from the lack of independence in the determination of the dependant variable, thus precluding statistical inferences, it is useful for quantifying the sensitivity of stomata to D (Oren et al. 1999), particularly for comparative purposes. All statistical analyses were conducted with SPSS (Version 10.0.6, SPSS, Chicago, IL) and Statistica software (Version 5.95, Tulsa, OK). 752 MARTÍNEZ-VILALTA ET AL. Figure 2. Difference in volumetric soil water content (θ0–25) between the drought-exposed and control plots. The beginning of the drought treatment is marked with an arrow. Shaded areas indicate the periods in which sap flow was monitored. Sap flow and canopy stomatal conductance in relation to environmental variables Leaf-related sap flow showed a similar temporal pattern for the three species studied in 1999 and for A. unedo in 2000 (at both the control and the drought-exposed lots). In all cases, maximum values were reached during early summer (DOY = 168–185, depending on the species and year) (Figure 5) when D started to increase and soil water content was still relatively high (θ0–25 > 0.23 m3 m –3) (Figures 1 and 5). Maximum Ql was similar for the three species in 1999 (P = 0.141 when comparing A. unedo and P. latifolia; Q. ilex could not be compared statistically because of the lack of replicates) (Figure 5). In the case of A. unedo and P. latifolia, this similarity was produced by similar sap flux densities and similar leaf-to-sapwood area ratios (Al:ASW) (Table 1). Instead, in the only Q. ilex individual measured, consistently higher sap flux density (data not Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 (P < 0.01). Predawn water potentials remained > –1.2 MPa, except during the summer months of both years when they reached minimum values ranging between –2.7 MPa (A. unedo control) and –6.2 MPa (P. latifolia) (Figure 3a). Midday water potentials were also lowest in P. latifolia (P < 0.01), but the difference between A. unedo and Q. ilex was not significant (P = 0.09) (Figure 3b). The difference between ΨMD and ΨPD decreased significantly during the summer in the three species (P < 0.01). The reduction was largest in Q. ilex (P < 0.05), which reached < 0.1 MPa at the peak of the 2000 drought. The slope of the linear regression of ΨMD against ΨPD was lower for Q. ilex (slope = 0.66; significantly < 1, P < 0.05) than for P. latifolia (0.92; not different than 1, P = 0.03) (Figure 4). The regression between ΨMD and ΨPD was not significant for A. unedo in the control plot. For Q. ilex and P. latifolia, water potentials tended to be slightly lower in 2000 than in 1999 during the summer (Figure 3). This result is consistent with slightly lower soil water contents in 2000 (Figure 1), but should be interpreted with caution because few days were measured each year. Figure 3. Seasonal patterns of (a) predawn leaf water potential, (b) midday leaf water potential and (c) whole-plant hydraulic conductance (kS-L) during 1999 and 2000. Arbutus unedo (Au) individuals were measured only in 2000, and are segregated between treatments (dr = drought and ct = control). Phillyrea latifolia (Pl) and Q. ilex (Qi) are not segregated because no difference was found between treatments. Bars represent standard errors. shown) was compensated by higher Al:ASW (Table 1). In A. unedo, maximum Ql was 112% higher in 2000 than in 1999 (P < 0.05) (Figure 5). During the summer, the continuous increase in SMD resulted in a reduction in Ql in all species (Figure 5). Vapor pressure deficit (and ET) showed no clear pattern during the same period (Figure 5). The sensitivity of Ql to SMD was larger in A. unedo than in P. latifolia (P < 0.05), and much larger in the only measured Q. ilex individual than in these other species (Figure 6a). Consistent with that result, estimated kS-L decreased sharply during the summer in the measured Q. ilex individual, whereas it remained approximately constant in P. latifolia (P > 0.8 when comparing early and late season val- TREE PHYSIOLOGY VOLUME 23, 2003 SAP FLOW OF THREE MEDITERRANEAN WOODY SPECIES 753 of the study period, Ql in this plot was reduced to values about 50% of those in the control plot. Consistent with this pattern, the reduction in Gs from early to late summer was larger in the dry plot than in the control plot for any value of D (Figure 8b). In agreement with a treatment effect, kS-L decreased significantly during summer 2000 in the drought-exposed plot (P = 0.02) but not in the control plot (P > 0.9) (Figure 3c). Discussion ues) (Figure 3c). At the end of the summer, Ql of P. latifolia was approximately twice that of A. unedo, and the difference was even larger in relation to the measured Q. ilex individual (Figure 5). The mean decline in Ql between early summer and the peak of the drought (SMD = 308 mm at the end of the summer) was 74.1% for A. unedo, 57.8% for P. latifolia and 87.1% for the Q. ilex individual measured. A similar pattern, with significantly lower Ql reductions in P. latiflolia than in the other species, was observed during the summer of 2000 in the drought-exposed plot (data not shown). However, these data must be interpreted with caution because they combine species and treatment effects, and as a result, have not been considered in our analysis. The relationship between Ql and SMD in A. unedo differed between years (P < 0.05), with a greater decrease in Ql with SMD in 2000 than in 1999 (Figure 6). To investigate the causes of the drought-induced decrease in Ql, Gs was derived from Ql. The relationship between Gs and D in the three species changed during the study period (Figure 7), indicating increased stomatal closure as the drought progressed. The model corresponding to late summer (higher SMD) was divided by the early summer model for each species to calculate the percent reduction in Gs at a given D as a result of soil moisture depletion. In 1999, the reduction in Gs from early to late summer ranged from < 50% in P. latifolia to > 80% in the measured Q. ilex individual, with A. unedo showing an intermediate reduction of about 60% (Figure 8a). Late season reduction in Gs of A. unedo was larger in 2000 than in 1999, reaching > 85% at high D (Figure 8b). The effects of the drought treatment were tested only in A. unedo, which was the only species that showed lower water potentials in response to the drought treatment in the year 2000. Despite similar maximum Ql in the control and drought-exposed plots (P > 0.4) (Figure 5), a treatment effect was apparent during the summer (P < 0.05, Figure 6b), with higher sensitivity to increased SMD in the dry plot. At the end TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Figure 4. Relationship between midday and predawn water potentials for A. unedo (Au), P. latifolia (Pl) and Q. ilex (Qi). Linear regression lines are depicted when significant, and the estimated regression slopes are shown in parentheses. The 1:1 relationship is also shown. Abbreviations: dr = drought; and ct = control. All studied trees showed similar patterns of seasonal water use, with maximum Ql in early summer and acute reductions as the summer drought progressed (Figure 5). This pattern is typical in Mediterranean evergreen forests (e.g., Tognetti et al. 1998, Infante et al. 2001). Our estimates of daily sap flows are consistent with published values for trees of the same species in similar forests. Teixeira et al. (1998), for example, measured sap flow in Q. ilex and A. unedo in an evergreen forest in southern France. Although they provided daily sap flow estimates instead of Ql or sap flux densities, their mean values from April to October (3.7 ± 0.4 kg day –1 for Q. ilex and 2.1 ± 0.4 kg day –1 for A. unedo) are similar to ours (2.7–4.1 kg day –1 for Q. ilex and 0.8–1.2 kg day –1 for A. unedo). The larger values for A. unedo are probably because their site was wetter than ours and their measured trees slightly larger. Our results imply that the sharp reduction in Ql in the studied trees as the drought progressed was caused by increased stomatal control over water loss. This is supported by the progressive reduction in estimated Gs during the summer (Figures 7 and 8) and by the low decoupling coefficients typical of Q. ilex woodlands (Infante et al. 1997) and suggested by our results (high ratio of boundary layer to stomatal conductance throughout the study period). Despite qualitatively similar patterns of seasonal water use, the magnitude of the decrease in sap flow during the summer and the seasonal dynamics of water potentials varied among species (Figures 3–6), suggesting that the responses to drought were also species-dependent. The results agree with our initial hypothesis that the effects of drought on sap flow would be highest in Q. ilex and lowest in P. latifolia, and with previous leaf-based measurements showing that A. unedo and Q. ilex close stomata at relatively high water potentials (Castell et al. 1994, de Lillis and Mirgone 1994, Sala and Tenhunen 1994, Peñuelas et al. 1998). The smaller slope of the relationship between ΨPD and ΨMD for Q. ilex compared with that for P. latifolia (Figure 4) further supports this interpretation. Data from the same study area show that A. unedo and Q. ilex are much more vulnerable to xylem embolism than P. latifolia (Table 1). Vulnerability to embolism and stomatal control are associated because plants must avoid dangerous losses of conductivity in the xylem caused by low water potentials (Oren et al. 1999, Sperry 2000; see below). We estimated native xylem embolism by combining the measurements of water potentials from this study with the vulnerability curves measured by Martínez-Vilalta et al. (2002b) in surface roots and stems of the three species studied and in the same area. 754 MARTÍNEZ-VILALTA ET AL. Predawn water potentials were assumed to provide an estimate of minimum xylem water potential in roots, whereas midday values were taken as a measure of xylem water potentials in stems. Estimates of xylem embolism increased sharply during the summer, reaching values > 40% in all species and tissues at the peak of the drought (Figure 9). Quercus ilex experienced greater drought-induced xylem embolism than P. latifolia and, if the pattern observed in 2000 holds for 1999, A. unedo as well (Figure 9). Xylem embolism was lowest in P. latifolia, even though water potentials were much lower in this species than in A. unedo or in Q. ilex. In general, resistance to xylem embolism is well adjusted to the range of water potentials within which plants operate (Hacke and Sperry 2001). The higher % embolism in Q. ilex is in agreement with the larger decrease in kS-L that we found in the Q. ilex individual that was measured during summer 1999 (Figure 3c). Under steady-state conditions, and assuming a high degree of aerodynamic coupling with the atmosphere, stomatal conductance (gs) is linked to D by the following relationship (Oren et al. 1999): g s = K L ∆Ψ/D (4) where KL is leaf-specific hydraulic conductivity, and ∆Ψ is the gradient of water potential from soil to leaves. Because xylem embolism reduces KL at low water potentials and ∆Ψ tends to decrease with decreasing soil water potential (e.g., Figure 4), a reduction of gs (or Gs) at any given D is expected as the soil dries (Figure 7). In agreement with our initial hypothesis, this reduction was greater in the only measured Q. ilex individual and smaller in P. latifolia (Figure 8), showing a high degree of correspondence with the values of native embolism estimated TREE PHYSIOLOGY VOLUME 23, 2003 Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Figure 5. Mean daytime vapor pressure deficit (D) and daily leaf-related sap flow (Ql) for each study species from early to late summer of 1999 and 2000. Only A. unedo individuals were monitored in 2000. The values are species means except for Q. ilex during summer 1999, when one individual was measured. Error bars represent standard errors (inter-individual variability). SAP FLOW OF THREE MEDITERRANEAN WOODY SPECIES 755 Figure 6. Relationship between daily leaf-related sap flow (Ql) (% relative to the maximum value reached during early summer) and soil moisture deficit (SMD) for A. unedo (Au), P. latifolia (Pl) and Q. ilex (Qi) during summer 1999 (a), and for A. unedo during summer 2000 (b). All regressions are highly significant (P < 0.001). Data points are species means, except for Q. ilex. Abbreviations: dr = drought; and ct = control. ences in the distribution of precipitation (Figure 1). Cumulative precipitation to the beginning of the summer was 289 mm in 1999 and 393 mm in 2000, and what is probably more important, in 2000, high rainfall and soil humidity coincided with a period of high evaporative demand (Figure 1). This suggests that interannual variability in sap flow can be substantial in these communities (about a twofold change in maximum Ql between 1999 and 2000 in A. unedo) and should be incorpo- Figure 7. Relationship between canopy stomatal conductance (Gs) and mean daytime vapor pressure deficit (D) for each species studied during 1999 and 2000. Three periods during the season are differentiated. Logarithmic regression curves are depicted for early and late season values. Values are species means, except for Q. ilex. TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 for each species (Figure 9). One advantage of this analysis is that differences among species in rooting extension and depth, which we have not studied, are partially incorporated via differences in ΨPD. In A. unedo, the dynamics of Ql varied between years and as a result of the drought treatment (Figures 5 and 6). Higher maximum Ql and a sharper decrease during the summer in 2000 were probably associated with between-years differ- 756 MARTÍNEZ-VILALTA ET AL. Figure 8. Reduction in canopy stomatal conductance (Gs) from early to late summer of 1999 (a) and 2000 (b) as a function of vapor pressure deficit (D), calculated from the ratio between the models fitted in Figure 7. Abbreviations: Au = A. unedo; Pl = P. latifolia; Qi = Q. ilex; dr = drought; and ct = control. Figure 9. Seasonal patterns of estimated percent loss of hydraulic conductivity caused by xylem embolism in roots (a) and stems (b). Data were obtained by combining leaf water potentials (Figure 3) and vulnerability curves (Martínez-Vilalta et al. 2002b). Abbreviations: Au = A. unedo; Pl = P. latifolia; Qi = Q. ilex; dr = drought; and ct = control. ciated with a 36% reduction in ΨPD (Figure 3), an increase in estimated % loss of xylem conductivity (Figure 9), a significant decrease in kS-L (Figure 3c) and a 50% reduction in Ql at the peak of the 2000 drought (Figure 6b). Although the treatment was not truly replicated in this study, the similarity of the course followed by soil water content, leaf water potential, gas exchange and tree growth among study plots suggests that they were representative of the whole system. Arbutus unedo was the only species in which a treatment effect was observed in 2000; consistent with sap flow results, water exclusion decreased growth by 66% and modified the carbon isotopic composition of leaves (R. Ogaya and M. Mangirón, unpublished results). In conclusion, our results suggest that the three species studied have contrasting strategies to deal with low water availability, as has been shown for other co-occurring Mediterranean species (e.g., Nardini et al. 1999). Although some of our conclusions must remain tentative because of small sample sizes, particularly in the case of Q. ilex, the following picture emerges. Low vulnerability to xylem embolism allows P. latifolia to maintain higher sap flow at lower water potentials than the other species. On the other hand, A. unedo and Q. ilex have greater stomatal control over water loss, thereby tending to avoid low water potentials that could cause a dangerous degree of xylem embolism (“runaway embolism,” Tyree and Sperry 1988). Under extremely dry conditions, however, stomatal control in these species was insufficient to prevent extensive loss of hydraulic conductivity as a result of embolism, particularly in Q. ilex. The prediction that, in relation to water stress, Q. ilex at our study site is close to its distributional limit is in agreement with previous ecophysiological (Lo Gullo and Salleo 1993, Peñuelas et al. 1998) and modeling (Martínez-Vilalta et al. 2002a) studies, and with the acute impact that the 1994 drought had on Q. ilex populations (Lloret and Siscart 1995, Peñuelas et al. 2000). Given the increase in aridity predicted by climate change models in the Mediterranean basin (IPCC 2001) and already observed in NE Spain (Piñol et al. 1998), these results should be taken into consideration when predicting the impact of climate change on vegetation. However, acclimation to water stress or to increased atmospheric CO2 concentration could compensate, at TREE PHYSIOLOGY VOLUME 23, 2003 Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 rated when representative values of stand transpiration are required. High interannual variability in evapotranspiration has been observed in the same study area at the catchment level (Piñol et al. 1991). Results for the treatment effect on A. unedo generally agreed with other through-fall manipulation experiments (Borghetti et al. 1998, Hanson et al. 2001 and references therein). The treatment caused a maximum decrease in the water content of surface soil of about 15%. This effect was asso- SAP FLOW OF THREE MEDITERRANEAN WOODY SPECIES least in part, for a decrease in water availability (Gebre et al. 1998, Tognetti et al. 1998, Osborne et al. 2000). Acknowledgments Field assistance by Jordi Sardans, Marta Mastrantonio and Nuri Cañellas is greatly appreciated. Dr. Pilar Llorens, Dr. Peter Becker and several anonymous reviewers provided valuable comments on an earlier version of the manuscript. We also thank the DARP (Generalitat de Catalunya) for allowing us to conduct research in the Poblet Forest. J.M. benefited from a FPI grant from the Catalan government. Financial support was obtained from the Spanish CLI970344 project to J. Peñuelas and J. Piñol. References IPCC. 2001. Climate change 2001: the scientific basis. Eds. J.T. Houghton, Y. Ding, D.J. Griggs, M. Noguer, P.J. van der Linden, X. Day, K. Maskell and C.A. Johnson. Cambridge University Press, Cambridge, 881 p. Jones, H.G. 1992. Plants and microclimate. 2nd Edn. Cambridge University Press, Cambridge, 429 p. Lledó, M.J. 1990. Compartimentos y flujos biogeoquímicos en una cuenca de encinar del Monte de Poblet. Ph.D. Thesis, Univ. d’Alacant, Spain, 292 p. Lloret, F. and D. Siscart. 1995. Los efectos demográficos de la sequía en poblaciones de encina. Cuadernos de la Sociedad Española de Ciéncias Forestales 2:77–81. Lo Gullo, M.A. and S. Salleo. 1993. Different vulnerabilities of Quercus ilex L. to freeze- and summer drought-induced xylem embolism: an ecological interpretation. Plant Cell Environ. 16: 511–519. Martínez-Vilalta, J.M, J. Piñol and K. Beven. 2002a. A hydraulic model to predict drought-induced mortality in woody plants: an application to climate change in the Mediterranean. Ecol. Model. 155:127–147. Martínez-Vilalta, J.M., E. Prat, I. Oliveras and J. Piñol. 2002b. Xylem hydraulic properties of roots and stems of nine Mediterranean woody species. Oecologia 133:19–29. Nardini, A., M.A. Lo Gullo and S. Salleo. 1999. Competitive strategies for water availability in two Mediterranean Quercus species. Plant Cell Environ. 22:109–116. Ogaya, R., J. Peñuelas, J. Martinez-Vilalta and M. Mangiron. 2003. Effects of drought on diameter increment of Quercus ilex, Phillyrea latifolia and Arbutus unedo in a holm oak forest of NE Spain. For. Ecol. Manage. In press. Oliveras, I. and P. Llorens. 2001. Medium-term sap flux monitoring in a Scots pine stand: analysis of the operability of the heat dissipation method for hydrological purposes. Tree Physiol. 21:473–480. Oren, R. and D.E. Pataki. 2001. Transpiration in response to variation in microclimate and soil moisture in southeastern deciduous forests. Oecologia 127:549–559. Oren, R., J.S. Sperry, G.G. Katul, D.E. Pataki, B.E. Ewers, N. Phillips and K.V.R. Schäfer. 1999. Survey and synthesis of intra- and interspecific variation in stomatal sensitivity to vapour pressure deficit. Plant Cell Environ. 22:1515–1526. Osborne, C.P., P.L. Mitchell, J.E. Sheehy and F.I. Woodward. 2000. Modelling the recent historical impacts of atmospheric CO2 and climate change on Mediterranean vegetation. Global Change Biol. 6:445–458. Pataki, D.E., R. Oren, G. Katul and J. Sigmon. 1998. Canopy conductance of Pinus taeda, Liquidambar styraciflua and Quercus phellos under varying atmospheric and soil water conditions. Tree Physiol. 18:307–315. Pataki, D.E., R. Oren and W.K. Smith. 2000. Sap flux of co-occurring species in a western subalpine forest during seasonal soil drought. Ecology 81:2557–2566. Peñuelas, J., I. Filella, J. Llusià, D. Siscart and J. Piñol. 1998. Comparative field study of spring and summer leaf gas exchange and photobiology of the Mediterranean trees Quercus ilex and Phillyrea latifolia. J. Exp. Bot. 49:229–238. Peñuelas, J., I. Filella, F. Lloret, J. Piñol and D. Siscart. 2000. Effects of a severe drought on water and nitrogen use by Quercus ilex and Phillyrea latifolia. Biol. Plant. 43:47–53. Phillips, N. and R. Oren. 1998. A comparison of daily representations of canopy conductance based on two conditional time-averaging methods and dependence of daily conductance on environmental factors. Ann. Sci. For. 55:217–235. TREE PHYSIOLOGY ONLINE at http://heronpublishing.com Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Allen, R.G., L.S. Pereira, D. Raes and M. Smith. 1998. Crop evapotranspiration. Guidelines for computing crop water requirements. FAO Irrigation and drainage paper 56, Rome, 290 p. Bolós, O. and J. Vigo. 1990–1995. Flora dels Països Catalans. Vol. 2–3. Barcino, Barcelona, 2152 p. Borghetti, M., S. Cinnirelle, F. Magnani and A. Saracino. 1998. Impact of long-term drought on xylem embolism and growth in Pinus halepensis Mill. Trees 12:187–195. Castell, C., J. Terradas and J.D. Tenhunen. 1994. Water relations, gas exchange, and growth of resprouts and mature plant shoots of Arbutus unedo L. and Quercus ilex L. Oecologia 98:201–211. Clearwater, M.J., F.C. Meinzer, J.L. Andrade, G. Goldstein and N.M. Holbrook. 1999. Potential errors in measurement of nonuniform sap flow using heat dissipation probes. Tree Physiol. 19:681–687. de Lillis, M. and E. Mirgone. 1994. Water relations of some Mediterranean evergreen sclerophylls: some considerations on sclerophylly and microphylly. Ecol. Mediterr. 20:39–49. Edwards, W.R.N., P. Becker and J. Èermák. 1996. A unified nomenclature for sap flow measurements. Tree Physiol. 17:65–67. Gebre, G.M., T.J. Tschaplinski and T.L. Shirshac. 1998. Water relations of several hardwood species in response to throughfall manipulation in an upland oak forest during a wet year. Tree Physiol. 18:299–305. Granier, A. 1985. Une nouvelle méthode pour la mesure du flux de sève brute dans le tronc des arbres. Ann. Sci. For. 42:193–200. Granier, A. 1987. Evaluation of transpiration in a Douglas-fir stand by means of sap flow measurements. Tree Physiol. 3:309–320. Granier, A., P. Biron, N. Bréda, J.-Y. Pontailler and B. Saugier. 1996. Transpiration of trees and forest stands: short and long-term monitoring using sap-flow methods. Global Change Biol. 2:265–274. Hacke, U.G. and J.S. Sperry. 2001. Functional and ecological xylem anatomy. Perspect. Plant Ecol. Evol. Syst. 4:97–115. Hanson, P.J., D.E. Todd and J.S. Amthor. 2001. A six-year study of sapling and large-tree growth and mortality responses to natural and induced variability in precipitation and throughfall. Tree Physiol. 21:345–358. Hereter, A. and J.R. Sánchez. 1999. Experimental areas of Prades and Montseny. In Ecology of Mediterranean Evergreen Oak Forests. Eds. F. Rodà, J. Retana, C.A. Gracia and J. Bellot. SpringerVerlag, Berlin, pp 15–27. Infante, J.M., S. Rambal and R. Joffre. 1997. Modelling transpiration in holm-oak savannah: scaling up from the leaf to the tree scale. Agric. For. Meteorol. 87:273–289. Infante, J.M., A. Mauchamp, R. Fernández-Alés, R. Joffre and S. Rambal. 2001. Within-tree variation in transpiration in isolated evergreen oak trees: evidence in support of the pipe model theory. Tree Physiol. 21:409–414. 757 758 MARTÍNEZ-VILALTA ET AL. Sperry, J.S. 2000. Hydraulic constraints on plant gas exchange. Agric. For. Meteorol. 104:13–23. Teixeira, J., C. Damesin, S. Rambal and R. Joffre. 1998. Retrieving leaf conductances from sap flows in a mixed Mediterranean woodland: a scaling exercise. Ann. Sci. For. 55:173–190. Terradas, J. 1999. Holm oak and holm oak forests: an introduction. In Ecology of Mediterranean Evergreen Oak Forests. Eds. F. Rodà, J. Retana, C.A. Gracia and J. Bellot. Springer-Verlag, Berlin, pp 3–14. Tognetti, R., A. Longobucco, F. Miglietta and A. Raschi. 1998. Transpiration and stomatal behavior of Quercus ilex plants during the summer in a Mediterranean carbon dioxide spring. Plant Cell Environ. 21:613–622. Tretiach, M. 1993. Photosynthesis and transpiration of evergreen Mediterranean and deciduous trees in an ecotone during a growing season. Acta Oecol. 14:341–360. Tyree, M.T. and J.S. Sperry. 1988. Do woody plants operate near the point of catastrophic xylem dysfunction caused by dynamic water stress? Plant Physiol. 88:574–580. Whitehead, D. and P.G. Jarvis. 1981. Coniferous forests and plantations. In Water Deficits and Plant Growth. Ed. T.T. Kozlowski. Academic Press, New York, pp 49–152. Wullschleger, S.D., F.C. Meinzer and R.A. Vertessy. 1998. A review of whole-plant water use studies in trees. Tree Physiol. 18: 499–512. TREE PHYSIOLOGY VOLUME 23, 2003 Downloaded from treephys.oxfordjournals.org at University of Portland on May 24, 2011 Piñol, J., M.J. Lledó and A. Escarré. 1991. Hydrological balance of two Mediterranean forested catchments (Prades, northeast Spain). Hydrol. Sci. J. 36:95–107. Piñol, J., J. Terradas and F. Lloret. 1998. Climate warming, wildfire hazard, and wildfire occurrence in coastal eastern Spain. Clim. Change 38:345–357. Potvin, C., M.J. Lechowicz and S. Tardif. 1990. The statistical analysis of ecophysiological response curves obtained from experiments involving repeated measures. Ecology 71:1389–1400. Rodà, F., X. Mayor, S. Sabaté and V. Diego. 1999. Water and nitrogen limitation to primary production. In Ecology of Mediterranean Evergreen Oak Forests. Eds. F. Rodà, J. Retana, C.A. Gracia and J. Bellot. Springer-Verlag, Berlin, pp 183–194. Sala, A. and J.D. Tenhunen. 1994. Site-specific water relations and stomatal response of Quercus ilex L. in a Mediterranean watershed. Tree Physiol. 14:601–617. Sala, A., S. Sabaté, C.A. Gracia and J.D. Tenhunen. 1994. Canopy structure within a Quercus ilex forested watershed: variations due to location, phenological development, and water availability. Trees 8:254–261. Schäfer, K.V.R., R. Oren and J.D. Tenhunen. 2000. The effect of tree height on crown level stomatal conductance. Plant Cell Environ. 23:365–375. Scholander, P.F., H.T. Hammel, E.D. Bradstreet and E.A. Hemmingsen. 1965. Sap pressure in vascular plants. Science 148: 339–346.