Tree Physiology 24, 901-909 ? 2004 Heron Publishing?Victoria, Canada Dynamics of transpiration, sap flow and use of stored water in tropical forest canopy trees FREDERICK C. MEINZER/'^ SHELLEY A. JAMES^nd GUILLERMO GOLDSTEIN'^ USDA Forest Service, Forestiy Sciences Laboratory, 3200 SW Jefferson Way, Corvallis, OR 97331, USA Corresponding author (fineinzer@fs.fed.us) Pacific Center for Molecular Biodiversity, Bishop Museum, Honolulu, HI 96817, USA Department of Biology, University of Miami, Coral Gables, FL 33124, USA Received October 23, 2003; accepted February 15, 2004; published online June 1, 2004 Summary In large trees, the daily onset of transpiration causes water to be withdrawn from internal storage compart- ments, resulting in lags between changes in transpiration and sap flow at the base of the tree. We measured time courses of sap flow, hydraulic resistance, plant water potential and sto- matal resistance in co-occurring tropical forest canopy trees with trunk diameters ranging from 0.34-0.98 m, to determine how total daily water use and daily reliance on stored water scaled with size. We also examined the effects of scale and tree hydraulic properties on apparent time constants for changes in transpiration and water flow in response to fluctuating environ- mental variables. Time constants for water movement were es- timated from whole-tree hydraulic resistance {K) and capaci- tance (C) using an electric circuit analogy, and from rates of change in water movement through intact trees. Total daily wa- ter use and reliance on stored water were strongly correlated with trunk diameter, independent of species. Although total daily withdrawal of water from internal storage increased with tree size, its relative contribution to the daily water budget (~10%) remained constant. Net withdrawal of water from stor- age ceased when upper branch water potential corresponded to the sapwood water potential (*Fsw) at which further withdrawal of water from sapwood would have caused *Ps^ to decline pre- cipitously. Stomatal coordination of vapor and liquid phase re- sistances played a key role in limiting stored water use to a nearly constant fraction of total daily water use. Time constants for changes in transpiration, estimated as the product of whole- tree R and C, were similar among individuals (~0.53 h), indi- cating that R and C co-varied with tree size in an inverse man- ner. Similarly, time constants estimated from rates of change in crown and basal sap flux were nearly identical among individu- als and therefore independent of tree size and species. Keywords: allometric relationships, hydraulic architecture, hydraulic capacitance, hydraulic resistance, scaling, time constants. Introduction In trees, reliance on stored water to temporarily replace tran- spirational losses is an important homeostatic mechanism, constraining leaf water deficits and maintaining photosyn- thetic gas exchange as increases in hydraulic path length in- crease hydraulic resistance with tree height (Goldstein et al. 1998, Phillips et al. 2003). Daily withdrawal of water from in- ternal storage compartments close to the sites of evaporative water loss transiently uncouples leaf water status from resis- tances associated with water movement through the soil, roots and portions of the stem (Meinzer 2002). Thus internal water storage confers elasticity to an otherwise inelastic hydraulic system. The relative contribution of stored water to daily tran- spiration has been estimated for a number of species and varies widely from as little as 10-20% (Loustau et al. 1996, Gold- stein et al. 1998, Kobayashi and Tanaka 2001, Maherali and DeLucia 2001) to as much as 30-50% (Waring and Running 1978, Holbrook and Sinclair 1992). Relative water storage ca- pacity is determined by the moisture release characteristics of the principal storage compartment tissues and components of the hydraulic architecture such as the leaf area:sapwood area ratio, provided the sapwood constitutes a significant storage compartment. Despite the increasing availability of estimates of water storage capacity in trees, relatively little is known about the daily dynamics of discharge and recharge of stored water and their consequences for stomatal regulation of gas exchange and leaf water status. In a few recent studies, Williams et al. (1996) attributed the afternoon decline in CO2 uptake in a Quercus-Acer stand to partial stomatal closure in response to depletion of stored water. Goldstein et al. (1998) showed that larger tropical trees with greater storage capacity maintained maximum rates of transpiration for a greater fraction of the day than smaller trees with a smaller storage capacity, and Phillips et al. (2003) showed that use of stored water in conif- erous and angiosperm species was concentrated in the morn- ing and early afternoon when conditions were most conducive for photosynthesis. However, the extent to which observed be- havior is species-specific or dependent in a universal manner on variables such as tree size, architecture and allometry, is unknown. 902 MEINZER, JAMES AND GOLDSTEIN The elasticity or capacitance associated with the release of stored water into the transpiration stream results in time lags between changes in transpiration and changes in liquid water movement through the xylem of leaves, stems and roots. In in- tact trees, use of stored water is often assessed from the lag be- tween changes in sap flow measured in the upper crown and near the base of the tree (Loustau et al. 1996, Goldstein et al. 1998). However, this approach, which assumes that sap now in upper branches is a surrogate for transpiration, does not ac- count for the water storage capacity of stem and leaf tissue dis- tal to the uppermost point where sap now is measured. Using an Ohm's Law analogy, time constants associated with lags between changes in environmental variables that affect tran- spiration and changes in water flow within the plant can be es- timated as the product of its hydraulic resistance and hydraulic capacitance (RC) between the soil and a reference point within the plant, usually the leaves. If either hydraulic resistance or capacitance varies throughout the day, the whole-tree time constant for environmentally-driven changes in water flow may be a dynamic rather than static property of water transport along the soil-to-atmosphere continuum. The objectives of the present study were to determine how total daily water use and daily reliance on stored water scaled with size among four co-occurring tropical forest canopy tree species, and to examine the effects of scale and tree hydraulic properties on apparent time constants for changes in water flow in response to fluctuations of the driving environmental variables. We employed a general electric circuit analogy to assess overall patterns of water use, water storage and apparent time constants among the four species studied rather than at- tempting to apply a specific RC model to describe their dy- namic behavior. We predicted that scaling of both water use and water storage capacity would be species-independent, and that compensatory variation in hydraulic resistance and capac- itance would result in similar time constants across a large range of tree sizes. Materials and methods Field site and plant material The study was carried out in a seasonally dry tropical forest in the Parque Natural Metropolitano, Panama City, Republic of Panama (09? 10' N, 79?51' E, elevation 50 m), at the site of the Smithsonian Tropical Research Institute canopy crane. Mean annual rainfall at the site is ~1800 mm, of which less than 150 mm falls during the dry season between January and April. One individual each of Cordia alliodora (Ruiz. & Pav.) Oken (Boraginaceae) and Schejflera morototoni (Aubl.) Mag- uire, Steyerm. & Frodin (Araliaceae) was studied during the dry season (February to April) of 2000, and one individual each of Anacardium excelsum (Bentero & Balb. ex Kunth) Skeels (Anacardiaceae) and Ficus ins?pida Willd. (Moraceae) was studied during the dry season of 2001 (Table 1). The mean maximum canopy height was ~35 m and numerous gaps were present, resulting in nearly complete exposure of the crowns of the study trees. The crane's gondola provided access to the crowns and upper trunks of the study trees. Sap flow, transpiration and water storage Variable length heat dissipation sap flow probes with a heated and reference sensor measuring length of 10 mm at the probe tip (James et al. 2002) determined sap flux at different radial depths and vertical positions in the four trees. Two replicate sets of probes were installed on opposite sides of the trunk near the base of each tree. Pairs of sensors were placed in an upward spiral around the trunk, 10 cm apart vertically and 5 cm apart circumferentially at five successive depths at a height of 3.1 m for A. excelsum, 3.5 m for F. ins?pida and 1.5 m for S. moro- totoni, and at four depths at 1.5 m height for C. alliodora. Sen- sor depths were 1.5, 4, 10, 17 and 24 cm for A. excelsum and F. ins?pida, 1.3, 3.6,7.6, 13.6 and 18 cm for S. morototoni, and 1.7,4, 7 and 11 cm for C. alliodora. A pair of sensors was also installed at a sapwood depth of approximately 1.3 cm in each of three (2001) or five (2000) replicate branches in the upper crown (branch diameter about 5 cm). For probe installation, two 38-gauge (2.58-mm-diameter) holes, separated axially by 10 cm, were drilled into the sap- wood. The sensors were coated with thermally conductive sili- cone heat sink compound prior to insertion. All probes were protected from direct sunlight and rainfall by reflective insula- tion and foam insulation in the branches. Concurrent differen- tial voltage measurements across the copper thermocouple leads were converted to a temperature difference between the heated and reference sensor (Ar). Signals from the sap flow probes were scanned every minute and 10-min means were re- corded with a data logger (CRIOX or CR21X, Campbell Sci- entific, Logan, UT) equipped with a 32-channel multiplexer (AM416, Campbell Scientific) and stored in a solid-state stor- age module (SMI92, Campbell Scientific). About 55-67 days of data were recorded for each tree, except the A. excelsum in- dividual for which intermittent sensor failure resulted in only Table 1. Characteristics of the individual representative trees studied. Whole-tree resistance values {R) were calculated from data presented in Meinzer et al. (2003). Using an electric circuit analogy, the time constant (T) is the product of resistance and capacitance (C) (Phillips et al. 1997). Species Diameter (m) He Anacardium excelsum 0.98 38 Ficus ins?pida 0.65 28 Schejflera morototoni 0.47 22 Cordia alliodora 0.34 26 ight (m) R(hMPakg-') C (kg MPa"') T(h) 0.0042 131.4 0.55 0.013 43.4 0.56 0.014 35.7 0.50 0.16 3.2 0.51 TREE PHYSIOLOGY VOLUME 24, 2004 DYNAMICS OF STORED WATER IN CANOPY TREES 903 22 days of data being recorded. The temperature difference between the heated and refer- ence sensors (AT) was converted to sap flux (v; g nT^ s"') based on the calibration of Granier (1985). The mass flow of sap corresponding to each trunk probe (F; g s"') was calcu- lated as: F = vA (1) where A (m^) is the cross-sectional area of the sapwood cal- culated as the ring area centered on the 10-mm-long sensor and extending to midway between two sensors of successive depth. The innermost sensor was considered to measure the sap flux to the estimated depth of heartwood as determined from wood cores (James et al. 2002, 2003). Sapwood area was calculated with the assumption of radial symmetry. Whole- tree daily water use (kg day"'), calculated from radial profiles of sap flow near the base of the tree, was assumed to be equal to total daily transpiration. Branch sap flow was used as a sur- rogate for whole-crown transpiration, which assumes that lags between change in rate of water vapor loss and change in branch sap flow were negligible compared with lags for chan- ge in basal sap flow. For each 10-min measurement, the mean sap flow of three to five branches was calculated and divided by the daily maximum to obtain an estimate of normalized whole-crown transpiration. Sap flow measured at the two out- ermost depths near the base of the trunk was normalized in a similar manner. Daily use of stored water for transpiration was estimated from lags between normalized crown and basal sap flow, as described by Goldstein et al. (1998). Positive values of crown minus basal sap flow indicate that water is being with- drawn from storage compartments located between the upper branches and the base of the trunk. Hydraulic resistance, stomatal resistance and capacitance Species-specific values of soil-to-terminal branch hydraulic resistance {K) were calculated from the inverse of hydraulic conductance data reported by Meinzer et al. (2003) for the same individuals. Determinations o?R were restricted to peri- ods when the lag between upper branch and basal sap flow was negligible indicating that the influence of capacitance on ap- parent resistance was minimal. In S. morototoni and C. allio- dora, leaf hydraulic resistance {Ri) was determined as: Therefore, A*FL should represent the transpiration-induced drop in A'?'L across the total hydraulic resistance of the leaf (Melcher et al. 1998). Leaf water potential was measured with a pressure chamber. Non-transpiring leaves were covered with aluminum foil and enclosed in plastic bags during the early evening preceding the measurement day. Stomatal resistance (^s) of S. morototoni and C. alliodora was measured with a steady state porometer (Model 1600, Li-Cor, Lincoln, NE) in order to determine whether there was an association between dynamic variation in r^ and AL- Whole-tree capacitance (kg MPa"') was determined as total daily withdrawal of water from storage, calculated from the cumulative differences between crown and basal sap flow (Goldstein et al. 1998), divided by the difference between *Pbr at the time withdrawal of water from storage ceased, and early morning values of 'i'br measured prior to the onset of sap flow. In addition, sapwood water release curves determined for the same individuals in a previous study (Meinzer et al. 2003) were used to estimate inflection points corresponding to the sapwood water potential (*Fsw) at which the capacitance, or slope of the curve (kg m"' MPa"'), changed from a nearly lin- ear phase to a distinctly nonlinear phase. Time constants and time lags Using an electric circuit analogy, the time constant (T) for changes in crown transpiration is the product of the soil-to-leaf hydraulic resistance and the total capacitance {RC ; Phillips et al. 1997). Whole-tree time constants (Table 1) were estimated by applying this model to the values of whole-tree capacitance and hydraulic resistance determined as described above. Time constants for changes in water flow were also estimated from time courses of sap flux in terminal branches and near the base of the main stem during the morning when irradiance and va- por pressure deficit were increasing steadily. The time con- stants for changes in crown and basal sap flux were estimated as the time required for sap flux to attain 63% of its daily maxi- mum steady state value. This procedure was expected to over- estimate the true time constants because environmental drivers of transpiration were changing in a continuous rather than stepwise fashion. Nevertheless, application of the same proce- dure to all of the trees was expected to reveal possible trends related to size and species. ?, A^P, (2) where A^PL is the difference in water potential between ex- posed transpiring leaves and covered non-transpiring leaves, and E is the corresponding transpiration rate per unit leaf area determined from values of upper branch sap flow normalized by the total leaf area distal to the sap flow probes. Branch tran- spiration was measured continuously (see above) and A*PL was measured at 1-2 h intervals throughout the morning and early afternoon (0800 to 1300 h). The water potential of covered leaves was assumed to be equivalent to branch water potential (^Pbr) at the point of leaf attachment (Begg and Turner 1970). Results Stored water use Total daily water use and stored water use for the four tropical tree species remained relatively constant during the dry sea- sons of 2000 and 2001 (Figure 1) with the exception of A. excelsum, in which total daily water use increased by about 20% (P < 0.01) during the study as a consequence of bud flushing and leaf expansion. Mean daily water use increased with tree size from 42 kg day"' in the 0.34-m-diameter C. alliodora tree to 785 kg day"' in the 0.98-m-diameter A. excelsum tree (Figure 2A). Daily withdrawal of water from storage followed a similar trend with increasing tree size (Fig- TREE PHYSIOLOGY ONLINE at http://heronpublishing.com 904 MEINZER, JAMES AND GOLDSTEIN ^ 180 '>, 160 03 "? 140 u> ^ 120 'T. 80 2000 2001 m 14 01 o 12 10 F o '*- b ? 4 03 ? n +-, ? .:'?-? ? I i I .- * ??? ?? ? ? 30 40 50 60 70 80 90 100 Day of the year 900 ? ? *? ? - 800 - ? ? ^^ - /OU ? " 600 - - 500 - - 400 - A - 300 -W* 'V V^?..^ ^^V'*^,^^. - 100 ? ? V ? ? - OU ? ?^ ?-? ? 60 - - 40 - ^ A - 20 \ '**' *A*6i?V ^A^^.'^.AI - 30 40 50 60 70 80 90 Day of the year Figure 1. Seasonal course of daily water use and use of stored water for Schejflera morototoni (I) and Cordia alliodora (?) during the 2000 dry season and Anacardium excelsum (?) and Ficus ins?pida (A) during the 2001 dry season. ure 2B), and the rankings of the trees according to total water use and stored water use were identical. Nevertheless, the rela- tive contribution of stored water to total daily transpiration re- mained nearly constant at about 10%, independent of total wa- ter use, tree size and species, as indicated by the slope of a highly significant (P < 0.01) linear regression fitted to the 800 7 400 03 200 S 100 B 50 CO 100 C. alliodora S. morototoni F. ins?pida A. excelsum 0.5 0.6 DBH(m) pooled seasonal data for all four trees (Figure 3). Daily courses of use and recharge of stored water determined from the differ- ence between crown and basal sap flow indicated that, in all four trees, stored water use increased abruptly shortly after sunrise, peaked at about 0800-0830 h, and ceased at 1000- 1100 h (Figure 4). Thus, about 10% of total daily water use had been withdrawn from storage by 1000 to 1100 h in all individ- uals, and varying amounts of recharge occurred during the re- mainder of the day and night (Figure 4, negative values). At least three independent lines of evidence suggest that none of the trees experienced progressive depletion of stored water, which would have resulted in increasing water deficits over the course of the dry season. First, no seasonal decline in total daily water use was observed in either year (Figure 1). Second, no seasonal increase in the nighttime maximum tem- os ?o 2 o ?) E o 03 g 03 100 80 1 ' 1 ? C. alliodora ? S. morototoni ? F. ins?pida - ? A. excelsum 1 1 1 1 1 60 - y^ ? 40 ty^ - 20 */^ y=0.10x-3.18 - ^f*^ r^ = 0.96 1 1 1 1 1 200 400 600 800 Water use (kg day~ ) 1000 Figure 2. Log-log plots of total daily water use and use of stored water in relation to tree diameter at breast height (DBH) for four tropical canopy trees. Values are means of 55-67 days of measurements. Stan- dard errors are smaller than symbols. Figure 3. Daily use of stored water in relation to total daily water use for four tropical canopy trees. The slope of the line is 0.10, indicating a contribution of 10% stored water to total daily transpiration, regard- less of tree size or species. TREE PHYSIOLOGY VOLUME 24, 2004 DYNAMICS OF STORED WATER IN CANOPY TREES 905 Q. CO CO C/3 CO Q. CO 12 14 Time (h) at which sapwood water release curves showed a transition from a nearly linear phase to a distinctly nonlinear phase ranged from -0.54 MPa in S. morototoni to -1.3 MPa in C. alliodora (Figure 5). When these values were plotted ag- ainst intact tree ^'f,^ values at the time daily use of stored water ceased (1000-1100 h), a linear relationship not significantly different from a 1:1 relationship was obtained (Figure 6). Thus about 10% of daily water use had been withdrawn from stor- age at the time ^^^r attained a value corresponding to the *Ps? at which further withdrawal of water from sapwood would have caused ^'^^ to decline precipitously (cf. Figures 5 and 6). Dynamics of transpiration and sap flow When upper crown sap flow was substituted for transpiration, the relative rates of the morning increase in transpiration for 2 days with similar time courses of solar radiation were found to be nearly identical among the four individuals studied (Fig- ure 7). Therefore, dynamic changes in transpiration in res- ponse to changes in the irradiance regime appeared to be inde- pendent of tree size, rates of whole-tree water use and water storage capacity. The preceding observation suggested that time constants for dynamic responses of transpiration to chan- ges in environmental variables such as irradiance were similar. To evaluate this possibility, whole-tree time constants were es- timated as the product of soil-to-branch hydraulic resistance obtained from an earlier study (Meinzer et al. 2003) and whole-tree capacitance calculated as described above (Ta- ble 1). Consistent with similar dynamic responses to changing irradiance (Figure 7), whole-tree time constants were nearly identical, ranging from 0.50-0.56 h. Relationships between tree size and the initial slope of the morning increase in sap flow near the base of the tree were strongly dependent on the scale at which they were assessed Figure 4. Representative time courses of the difference between crown and basal sap flow normalized with respect to their daily maxi- mum values. Positive values represent withdrawal of water from inter- nal storage and negative values represent recharge of storage compart- ments. perature difference between heated and reference sap flow sensors was observed (data not shown). If sapwood water con- tent had declined, sensor heat dissipation would have become slower, leading to seasonal increases in the temperature differ- ence between sensors when flow was at or near zero. Third, the sums of 24-h time courses of the difference between crown and basal sap flow were never significantly different from zero, consistent with no net withdrawal of water from storage over 24-h cycles. To identify potential regulatory mechanisms contributing to the similar timing of stored water use and relative reliance on stored water among the four species representatives, sapwood water potentials ('i'sw) from sapwood water release curves de- termined in a previous study (Meinzer et al. 2003) were com- pared with branch water potentials (*Pbi) corresponding to the times at which daily use of stored water ceased. Values of ^Ps? o m CO ? 300 S. morototoni 200 ._____^ ... C. alliodora \ - 100 - ^''^^^'>-- .^\ - 0 . ^T'AN -2.5 -2.0 -1.5 -1.0 %w (MPa) -0.5 0.0 Figure 5. Sapwood water release curves for S. morototoni and C. al- liodora. The dotted lines are linear regressions fitted to the nearly lin- ear portions of the curves and the dashed lines indicate inflection points at which the relationship between water released and sapwood water potential (4's^), determined psychrometrically, becomes dis- tinctly nonlinear. Water release curves for A. excelsum and F. ins?pida were intermediate between those of C. alliodora and S. morototoni. Inflection points were -1.3 MPa for C. alliodora, -0.90 MPa for F. in- s?pida, -0.70 MPa for A. excelsum and -0.55 MPa for S. morototoni. Data were obtained from Meinzer et al. (2003). TREE PHYSIOLOGY ONLINE at http://heronpublishing.com 906 MEINZER, JAMES AND GOLDSTEIN Figure 6. Relationship between brancli water potential (4'br) at the time net daily withdrawal from internal storage ceased and sapwood water potential (4'sw) corresponding to inflection points on sapwood water release curves (see Figure 5). The solid regression line did not differ significantly from a 1:1 relationship (dashed line). (Figure 8). The relative rate of increase in basal sap flow de- creased with increasing tree size (Figure 8A). In contrast to normalized flow, the rate of increase in whole-tree water use was greatest for larger trees (Figure 8B). Thus, despite the slower relative increase of sap flow in larger trees, the absolute increase was greater. However, the rate of increase in sap flux averaged over the entire basal sapwood area to remove the in- fluence of differences in total sapwood area among trees, was 1.0 ? C. alliodora ? *?* ^ ~ ? S. morototoni t!-" ? o 0.8 - ? F. ins?pida X o - ? A. excelsum to 0.6 ? ?. o ? N ? A (0 F 0.4 - I - o ? ^ 0.2 ' ^ Mi,', 1 lili - Day 51, 2000 1000 - ? - Day 49, 2000 / y" .? / / m 800 - F o 6UU - y^ b ^ ^ ?^ 3. 400 - ' . ' - II y'^y' n Q- 2?? n ^ 1 , 1 6.5 7.0 7.5 8.0 Time (h) 8.5 9.0 Figure 7. Time courses of normalized crown (upper branch) sap flow on days with similar morning time courses of solar radiation (PPF) from the 2 years of measurement. Q. ns en 0) en (0 0) 0.7 0.6 " 0.5 > 2 0.4 0.3 30 25 ^20 & 15 10 5 0 25 ^?20 CM 'E 10 5 ? C. ailiodora ? S. morototoni ? F. insipida # A. excelsum r =0.99 ?^ \ ' h r = 0.01 0.0 0.2 0.4 0.6 DBH(m) 0.8 1.0 Figure 8. Rates of the initial morning increase in sap flow (-0700 to 0900 h) calculated with (A) flow normalized with respect to the daily maximum, (B) whole-tree water use and (C) sap flux per unit of sap- wood area. similar among the four individuals (Figure 8C), implying that the time constants for changes in basal sap flux were similar among all individuals and independent of tree size or wood properties. Because the slopes of the initial morning increase in sap flow were similar among individuals when sap flow was ex- pressed per unit of sapwood area (sap flux), mean time courses of crown and basal sap flux were calculated to examine the characteristics of the lag between changes in crown and basal sap flow (Figure 9). The morning lag between crown transpi- ration and basal sap flow increased from about 0.25 h shortly after sunrise to about 0.75 h when crown transpiration first at- tained its maximum value as stomata began to restrict transpi- ration. The increasing lag time was indicative of the differ- ences in time constants for changes in crown and basal sap flow. Once stomata began to restrict transpiration, the lag quickly diminished and had disappeared by 0900 h (Figure 9). As both irradiance and atmospheric saturation deficit increas- ed in the morning, stomatal limitation of transpiration was as- sociated with increasing leaf hydraulic resistance (R?J during the same time period. Data obtained for C. alliodora and S. morototoni showed that stomatal resistance (r?) more than TREE PHYSIOLOGY VOLUME 24, 2004 DYNAMICS OF STORED WATER IN CANOPY TREES 907 CO CO Time (h) Figure 9. Mean time courses of crown and basal sap flux on represen- tative clear days for A. excelsum, F. ins?pida, C. alliodora and S. moro- totoni. Crown sap flux was obtained by normalizing mean upper branch sap flux with respect to its maximum value for the day, then us- ing the dimensionless values and corresponding values of total daily water use, determined from the basal sap flow probes, to calculate the time course of whole-crown transpiration (g s"'), which was normal- ized by basal sapwood area (m ). This procedure allowed both fluxes to be expressed at the same time scale. doubled between 0800 and 1300 h as Ri^ increased by an order of magnitude (Figure 10). Discussion Plant size played a dominant role in determining the water use and water storage characteristics of the four individuals stud- ied (Figure 2). Although total daily withdrawal of water from internal storage increased with tree size, and therefore total daily water use increased, the relative contribution of stored water to the daily water budget remained constant at about 10% (Figure 3). Similarly, Maherali and DeLucia (2001) re- 5.0 4.5 4.0 ? S. morototoni ? C. alliodora E 3.5 ?, 3.0 2.5 2.0 r =0.83 1.5 0.0 0.5 1.0 1.5 2.0 RL (MPa s m mmof ) 2.5 Figure 10. Relationship between stomatal resistance (?-J and leaf hy- draulic resistance (AL) fof C. alliodora and S. morototoni. Data were pooled from 2-3 days of measurements for each species. ported little site-specific and seasonal variation in relative water storage capacity of ponderosa pine (Pinus ponderosa Dougl. ex P. Laws. & C. Laws.) trees growing in desert and montane environments. In contrast, Phillips et al. (2003) re- ported that daily reliance on stored water increased with tree size in two temperate coniferous and one temperate angio- sperm species. Nevertheless, the relative reliance on stored water reported here and by Phillips et al. (2003) is within the range of estimates obtained in earlier studies (Loustau et al. 1996, Kobayashi and Tanaka 2001). The mechanisms that govern the extent to which transient withdrawal of water from internal storage is relied upon to re- place daily transpirational losses are uncertain. However, in the present study, a relationship was found between the timing of stored water use (Figure 4), sapwood water release charac- teristics (Figure 5), and values of *Pbi. at which net withdrawal of water from storage ceased (Figure 6). By the time approxi- mately 10% of the total daily water use had been withdrawn from storage, ^^r had fallen to values that corresponded to in- flection points on sapwood water release curves where ^F^^ began to decline precipitously with further withdrawal of wa- ter (cf. Figures 5 and 6). Although these threshold values of ^Psw were substantially less negative than values of *Pbr associ- ated with 50% loss of hydraulic conductivity from embolism, which ranged from -1.6 MPa in A. excelsum to -3.0 MPa in C. alliodora (Meinzer et al. 2003), it is possible that conserva- tive stomatal regulation of *Fbr at a setpoint corresponding to the point of "diminishing returns" on the sapwood water re- lease curve dampened daily fluctuations in the fraction of dys- functional xylem in stems. Stomata thus played a key role in limiting use of stored wa- ter to a nearly constant fraction of total daily water use inde- pendent of species and tree size. Rapid increases in r^ from early morning minimum values at 0800 h (data not shown) strongly limited transpiration, causing crown and basal sap flux to converge (Figure 9), thereby preventing further with- drawal of water from storage (Figure 4). Rapid increases in r^ during the morning were associated with corresponding in- creases in AL (Figure 10). Marked diel fluctuations in /?L have been noted in earlier studies (Zwieniecki et al. 2000, Bucci et al. 2003) and can be expected to strongly influence stomatal regulation of leaf water potential, and therefore *Pbr, because ^L is often the dominant resistance in the soil-to-leaf pathway (Meinzer 2002). Recent work has shown that diel fluctuations in /?L reflect diel cycles of embolism and refilling of xylem conduits (Bucci et al. 2003, Nardini et al. 2003). Thus, it ap- pears that the dynamics of embolism formation and repair in the terminal portions of the water transport pathway (i.e., leaves) may govern stomatal regulation of the water status of woody transport tissue to which the leaves are attached. In other words, dynamic responses of stomata to daily fluctua- tions in leaf embolism and hydraulic resistance suggest that a sensitive regulatory system exists that provides early warning of changes in evaporative demand. Failure of stomata to "an- ticipate" rapid increases in transpiration could result in sharp increases in stem xylem tension and loss of conductivity, espe- cially if the capability for buffering through capacitive dis- charge of water from sapwood storage has already been ex- TREE PHYSIOLOGY ONLINE at http://heronpublishing.com 808 MEINZER, JAMES AND GOLDSTEIN hausted. Moreover, reliance on transient xylem dysfunction in leaves to regulate the water status and degree of embolism in stems could be advantageous if embolism repair processes in woody stems are not as vigorous as those observed in leaves (Hacke and Sperry 2003). The data presented here suggest that time constants for changes in transpiration and sap flow in response to changing environmental conditions were similar among individuals of four tropical forest canopy tree species comprising a substan- tial range of tree sizes. Similar time constants for changes in transpiration estimated from sap flow in upper branches (Fig- ures 7 and 9) were associated with an inverse relationship be- tween whole-tree hydraulic resistance and capacitance (Ta- ble 1 and Meinzer et al. 2003). Whole-tree hydraulic resis- tance and capacitance thus appeared to co-vary in a predictable manner that resulted in their product (RC), an estimate of the time constant, remaining similar and therefore independent of tree size as proposed earlier (Hunt et al. 1991, Phillips et al. 1997, 1999). The mean time constant of 0.53 h for changes in transpiration (Table 1) is consistent with estimates of 0.5 to 0.7 h reported for two Pinus species (Phillips et al. 1997). An independent estimate of the mean time constant for the four study trees based on the linear portion of the increase in crown sap flux shown in Figure 9, and assuming a maximum sap flux of 30 g m"^ s"', yielded a value of about 0.7 h. Although this value was within the 0.5-0.7 h range cited above, it was about 30% greater than the estimate of 0.53 h obtained from Table 1. Step changes in environmental drivers of transpiration were implicitly assumed in calculating the time constants in Table 1, whereas slower, continuous changes in environmental vari- ables, namely irradiance and vapor pressure deficit, led to the behavior depicted in Figure 9. A key assumption involved in estimating both the amount of water withdrawn daily from internal storage and time con- stants for changes in transpiration was that water storage ca- pacity distal to the locations of branch sap flow measurements was negligible. Significant daily discharge and recharge of storage tissues distal to the uppermost sap flow sensors would have resulted in additional undetected lags between changes in transpiration and sap flow, and therefore underestimates of to- tal daily storage capacity and overestimates of time constants for changes in transpiration. However, technical and logistical constraints associated with sampling an adequate fraction of the crown prevented measurements of water flux in the small- est branches. Although porometric measurements of stomatal conductance can be used to measure leaf transpiration inside ventilated chambers, previous studies have shown that they of- ten provide unreliable estimates of transpiration from unen- closed leaves because the impact of leaf and canopy boundary layers on transpiration are ignored (Meinzer et al. 1995,1997). Apparent time constants for changes in sap flow near the base of the tree were strongly dependent on how the data were normalized and the scale at which they were expressed (Fig- ure 8). Normalizing sap flow with respect to its daily maxi- mum value yielded a negative relationship between tree size and the relative rate of change in sap flow (Figure 8 A), where- as, the rate of change in whole-tree sap flow increased sharply with increasing tree size (Figure 8B). However, when mean sap flow per unit of basal sap wood area was considered, its rate of change was independent of tree size and nearly constant (Figure 8C), suggesting uniform time constants for changes in basal sap flux among the four trees. The pattern in Figure 8C was consistent with the mean time course of basal sap flux shown in Figure 9. Conclusions Homeostatic mechanisms involving compensating adjust- ments in whole-tree capacitance and hydraulic properties over a range of tree sizes appeared to contribute to the stability of time constants for responses of transpiration and sap flow to changes in environmental variables. Additional homeostatic mechanisms were involved in the stomatal regulation of tree water status that limited withdrawal of water from internal storage compartments to about 10% of total daily water use in- dependent of tree size and species. The four individuals stud- ied showed a common relationship between tree size and total daily water use. Acknowledgments This research was supported by National Science Foundation Grant IBN 99-05012 to F. Meinzer and G. Goldstein. We thank the Smith- sonian Tropical Research Institute for providing facilities, logistical support and the expertise of the canopy crane operators. References Andrade, J.L., F.C. Meinzer, G. Goldstein, N.M. Holbrook, J. Cave- lier, R Jackson and K. Silvera. 1998. Regulation of water flux through trunks, branches and leaves in trees of a lowland tropical forest. Oecologia 115:463-471. Begg, J.E. and N.C. Turner. 1970. Water potential gradients in field tobacco. Plant Physiol. 46:343-346. Bucci, S.J., E.G. Scholz, G. Goldstein, EC. Meinzer and L. Stern- berg. 2003. Dynamic changes in hydraulic conductivity in petioles of two savanna tree species: factors and mechanisms contribut- ing to the refilling of embolized vessels. Plant Cell Environ. 26: 1633-1645. Goldstein, G., J.L. Andrade, EC. Meinzer, N.M. Holbrook, J. Cave- lier, P. Jackson and A. Celis. 1998. Stem water storage and diurnal patterns of water use in tropical forest canopy trees. Plant Cell En- viron. 21:397-406. Granier, A. 1985. Une nouvelle m?thode pour la mesure du flux de s?ve brute dans le tronc des arbres. Ann. Sei. Eor. 42:193-200. Hacke, U.G. and J.S. Sperry. 2003. Limits to xylem reflUing under negative pressure in Laurus nobilis and Acer negundo. Plant Cell Environ. 26:303-311. Holbrook N.M. and T.R. Sinclair. 1992. Water balance in the arbor- escent palm, Sabalpalmetto. II. Transpiration and stem water stor- age. Plant Cell Environ. 15:401-409. Hunt, E.R., S.W. Running and C.A. Eederer. 1991. Extrapolating plant water flow resistances and capacitances to regional scales. Agrie. Eor. Meteorol. 54:169-195. James, S.A., M.J. Clearwater, EC Meinzer and G. Goldstein. 2002. Heat dissipation sensors of variable length for the measurement of sap flow in trees with deep sapwood. Tree Physiol. 22:277-283. TREE PHYSIOLOGY VOLUME 24, 2004 DYNAMICS OF STORED WATER IN CANOPY TREES 809 James, S.A., F.C. Meinzer, G. Goldstein, D. Woodruff, T Jones, T Restom, M. Mejia, M. Clearwater and R Campanello. 2003. Ax- ial and radial water transport and internal water storage in tropical forest canopy trees. Oecologia 134:37-45. Kobayashi, Y. and T Tanaka. 2001. Water flow and hydraulic charac- teristics of Japanese red pine and oak trees. Hydrol. Proc. 15: 1731-1750. Loustau, D., P. Berbiger, P. Roumagnac, C. Arruda-Pacheco, J.S. Da- vid, M.J. Ferreira, J.S. Pereira and R. Tavares. 1996. Transpiration of a 64-year-old maritime pine stand in Portugal. 1. Seasonal course of water flux through maritime pine. Oecologia 107:33-42. Maherali, H. and E.H. DeLucia. 2001. Influence of climate-driven shifts in biomass allocation on water transport and storage in pon- derosa pine. Oecologia 129:481-491. Meinzer, FC. 2002. Co-ordination of liquid and vapor phase water transport properties in plants. Plant Cell Environ. 25:265-274. Meinzer, EC, G. Goldstein, P. Jackson, N.M. Holbrook, M.V. Gutier- rez and J. Cavelier. 1995. Environmental and physiological regula- tion of transpiration in tropical forest gap species: the influence of boundary layer and hydraulic properties. Oecologia 101: 514-522. Meinzer, EC, J.L. Andrade, G. Goldstein, N.M. Holbrook, J. Cave- lier and P. Jackson. 1997. Control of transpiration from the upper canopy of a tropical forest: the role of stomatal, boundary layer and hydraulic architecture components. Plant Cell Environ. 20: 1242-1252. Meinzer, EC, S.A. James, G. Goldstein and D. Woodruff. 2003. Whole-tree water transport scales with sapwood biophysical pro- perties in tropical forest canopy trees. Plant Cell Environ. 26: 1147-1155. Melcher, P.J., EC Meinzer, D.E. Yount, G. Goldstein and U. Zimmer- mann. 1998. Comparative measurements of xylem pressure in tran- spiring and non-transpiring leaves by means of the pressure cham- ber and the xylem pressure probe. J. Exp. Bot. 49:1757-1760. Nardini, A., S. Salleo and F Raimondo. 2003. Changes in leaf hy- draulic conductance correlate with leaf vein embolism in Cercis siliquastrum L. Trees 17:529?534. Phillips, N., A. Nagchaudhuri, R. Oren and G.G. Katul. 1997. Time constant for water uptake in loblolly pine trees estimated from time series of stem sap flow and evaporative demand. Trees 11 : 412-419. Phillips, N., R. Oren, R. Zimmermann and S.J. Wright. 1999. Tempo- ral patterns of water flux in trees and lianas in a Panamanian moist forest. Trees 14:116-123. Phillips, N.A., M.G. Ryan, B.J. Bond, N.G. McDowell, T.M. Hinck- ley and J. Cerm?k. 2003. Reliance on stored water increases with tree size in three species in the Pacific Northwest. Tree Physiol. 23: 237-245. Waring R.H. and S.W. Running. 1978. Sapwood water storage: its contribution to transpiration and effect upon the water conductance through the stems of old-growth Douglas-fir. Plant Cell Environ. 1:131-140. Williams, M., E.B. Rastetter, D.N. Eernandes et al. 1996. Modelling the soil-plant-atmosphere continuum in a Quercus-Acer stand at Harvard Forest: the regulation of stomatal conductance by light, ni- trogen and soil/plant hydraulic properties. Plant Cell Environ. 19: 911-927. Zwieniecki, M.A., L. Hutyra, M.V. Thompson and N.M. Holbrook. 2000. Dynamic changes in petiole specific conductivity in red ma- ple {Acer rubrum L.), tulip tree {Liriodendron tulipifera L.) and northern fox grape {Vitis labrusca L.). Plant Cell Environ. 23: 407-414. TREE PHYSIOLOGY ONLINE at http://heronpublishing.com