Stem water storage and diurnal patterns of water use in tropical forest canopy trees


Frederick C. Meinzer Fax: (808) 486 5020; e-mail:


Stem water storage capacity and diurnal patterns of water use were studied in five canopy trees of a seasonal tropical forest in Panama. Sap flow was measured simultaneously at the top and at the base of each tree using constant energy input thermal probes inserted in the sapwood. The daily stem storage capacity was calculated by comparing the diurnal patterns of basal and crown sap flow. The amount of water withdrawn from storage and subsequently replaced daily ranged from 4 kg d–1 in a 0·20-m-diameter individual of Cecropia longipes to 54 kg d–1 in a 1·02-m-diameter individual of Anacardium excelsum, representing 9–15% of the total daily water loss, respectively. Ficus insipida, Luehea seemannii and Spondias mombin had intermediate diurnal water storage capacities. Trees with greater storage capacity maintained maximum rates of transpiration for a substantially longer fraction of the day than trees with smaller water storage capacity. All five trees conformed to a common linear relationship between diurnal storage capacity and basal sapwood area, suggesting that this relationship was species-independent and size-specific for trees at the study site. According to this relationship there was an increment of 10 kg of diurnal water storage capacity for every 0·1 m2 increase in basal sapwood area. The diurnal withdrawal of water from, and refill of, internal stores was a dynamic process, tightly coupled to fluctuations in environmental conditions. The variations in basal and crown sap flow were more synchronized after 1100 h when internal reserves were mostly depleted. Stem water storage may partially compensate for increases in axial hydraulic resistance with tree size and thus play an important role in regulating the water status of leaves exposed to the large diurnal variations in evaporative demand that occur in the upper canopy of seasonal lowland tropical forests.


The water transport tissues of tall woody plants impose intrinsic limitations on the immediate replacement of transpirational water loss when the main water source is located in the soil. Because sap flow velocities in trees are of the order of 0·1–5 m h–1 (Granier 1987; Dye et al. 1996; Zang et al. 1996), several days may be required for the soil water to reach the upper crown of a 40-m-tall tree. In addition, the series of hydraulic resistances associated with moving water through xylem conduits may impose water deficits on leaves of large canopy trees during periods of high evaporative demand, even when soil water availability is high. One way of minimizing temporal imbalances in water supply and demand in tall trees is to utilize water stored within stem tissues that are closer to the sites of evaporative water loss than the soil reservoir.

Despite the potential importance of stem water storage in regulating the water economy of canopy trees, limited information exists on the contribution of internal water storage to their total daily water consumption, particularly in tropical trees. Tyree & Yang (1990) estimated that water withdrawn from living cells in stems of Thuja occidentalis contributed about 6% to the total daily transpirational water loss. On the other hand, Schulze et al. (1985) suggested that there was little water available in the main trunk of a Larix decidua tree, while stored water in branches contributed 24% to the total daily transpiration. The contribution of the internal water storage to daily transpirational losses apparently is not a fixed parameter in the water budget of trees. For example, the amount of water stored in the trunk of Pinus pinaster accounted for 12% of the daily transpiration when soil water was abundant, but increased to 25% of the daily transpiration at the end of summer following a period of drought (Loustau et al. 1996). Other estimates of the magnitude of sapwood water storage in temperate conifers during drought periods range as high as 50% of the water transpired over periods of several days (Waring & Running 1978; Waring, Whitehead & Jarvis 1979). Even if the amount of water obtained from internal storage during a day is relatively small, its role in the maintenance of a favorable leaf water and carbon balance and integrity of the water transport pathway could be important if it influences diurnal patterns of stomatal opening.

The number of studies on the role of internal water storage in determining diurnal patterns of water use in tropical canopy trees has been limited in part by the lack of simple, robust technology for measuring volumetric water fluxes in large stems and by the difficulties in gaining direct access to the upper part of the forest canopy. Relatively inexpensive constant energy input thermal sap flow probes (Granier 1987) have permitted the automatic recording of diurnal variations in sap flux in large trees. At the same time, large construction cranes have recently been adapted by equipping them with gondolas to gain access to the uppermost levels of forest canopies. Access to the upper part of forest canopies is allowing the detailed study of many physiological and ecological processes that were impossible to investigate just a few years ago.

The purpose of this study was to determine the contribution of stem water storage to the total daily transpirational losses, and the effect of the stem water storage on the diurnal pattern of water use in individuals of five canopy species in a seasonal tropical moist forest in Panama. Utilization of water from internal reserves was determined by comparing diurnal courses of sap flow measured in the upper crown and at the base of the tree. Individuals representing a large range of size and successional status were selected for study.


Field site and plant material

A construction crane was used to gain access to five canopy trees growing in the Parque Natural Metropolitano, a remnant of natural forest near Panama City, Panama (lat. 8°58’ N, long. 79°34’ W., alt. 50 m). The gondola of the crane was attached with cables to a rotating jib, and could be manoeuvred both horizontally and vertically with precision. The forest contained a mixture of evergreen, briefly deciduous, and a few dry season-deciduous species. Three evergreen trees (Ficus insipida Will, Moraceae, Luehea seemannii Tr & Planch, Tiliaceae, and Cecropia longipes Pitt., Moraceae) one brevideciduous tree [Anacardium excelsum, (Bertero & Balb.) Skeels, Anacardiaceae] and one facultatively deciduous tree (Spondias mombin L., Anacardiaceae) were chosen for this study. All of them were mid to upper canopy trees with good access from the crane. The height of the trees ranged from 35 m (A. excelsum) to 18 m (C. longipes) (Table 1). The mean annual rainfall at the site is about 1800 mm, of which less than 150 mm normally falls during the dry season between January and April. Measurements were made from January to April 1996. The total rainfall for the period was 377 mm, about 250% above normal.

Table 1.  . Characteristics of the five study trees. Values for sap flow rates, daily water storage, leaf water potential, and soil-to-leaf hydraulic conductance are means ± SE (sap flow and water storage n=6 to 8; leaf water potential and hydraulic conductance n=5 to 7) Thumbnail image of

Sap flow and micrometeorological measurements

Sap flow was measured with pairs of constant energy input thermal probes (UPGmbH, Munich, Germany) inserted in the sapwood of the stems. The upper probe of each pair was continuously heated and the lower unheated probe (located ≈ 15 cm upstream from the heated probe) gave the reference temperature of the sapwood. The temperature difference between the two probes is inversely related to sap flow density (Granier 1987). The 2-mm-diameter probes were inserted 20 mm into the sapwood. Bark was partially removed at the points of probe insertion to expose the outer surface of the sapwood. Probe temperature differences were measured every 10 s and averaged every 10 min using a datalogger (CR21X, Campbell Scientific Corp., Logan, UT, USA) and a 32-channel multiplexer (AM416-Campbell Scientific). Measurements were made simultaneously in three to four exposed branches in the upper crown and on opposite sides of the base of each tree during 10–14 consecutive days. The two sap flow measurements at the base of the same tree were typically within 2–20% of each other. Technical constraints associated with availability of instrumentation and manoeuvring the gondola of the crane precluded concurrent measurements in more than one tree at a time. The portion of the stem containing the probes was thermally insulated with foam and an outer layer of reflective material in order to minimize temperature gradients caused by radiant heating of the stem.

Sapwood area at the base of the tree was determined by injecting dye into the bole through holes made by a 5-mm-increment borer. After 1–2 h, a core was taken 2–4 cm above each dye injection point and the area of conducting tissue was determined from the thickness of the wood coloured by the dye as it moved up in the transpiration stream. Preliminary measurements using cores taken about 10–20 cm above the dye injection point yielded similar results but several cores were often required to locate the dye. The shorter distance between dye injection and core extraction was used to minimize damage to the trees. Dye injection in the upper branches 4–6 cm in diameter indicated that all of their xylem consisted of sapwood. Mass flow of sap was obtained by multiplying sap flow density by the sapwood cross-sectional area.

With the exception of A. excelsum, the crowns of the study trees were relatively shallow and well represented by the three to four upper branches selected for sap flow measurements. Although the crown of A. excelsum spanned a vertical distance of at least 20 m, previous measurements in this emergent, relatively isolated tree indicated that the magnitude and diurnal course of sap flow in the upper and lower crown were similar (Meinzer et al. 1993).

Photosynthetic photon flux density (PPFD), relative humidity and air temperature were monitored continuously near the crane in the upper canopy (about 25 m height) with a quantum sensor (LI-COR Inc., Lincoln, NE, USA), and a relative humidity and temperature probe (HMP35C, Campbell Scientific, Logan, UT, USA), respectively, connected to a second datalogger. Leaf temperature was measured with fine wire (0·08 mm) copper-constantan thermocouples in six leaves on each of the four branches used for sap flow measurements. The thermocouple junctions were affixed to the abaxial leaf surfaces, and temperature was continuously monitored with a datalogger. The vapour pressure difference between the leaf interior and the bulk air (VPD) was calculated from the saturation vapour pressure at leaf temperature and the ambient vapour pressure. The atmospheric saturation deficit (ASD) was calculated as the difference between saturation vapour pressure at the air temperature and ambient vapour pressure. Leaf water potential was measured psychrometrically. Leaf discs were obtained from attached leaves with a cork borer and sealed inside psychrometer chambers (81 series, JRD Merrill Specialty Equipment, Logan, Utah, USA) in the field. The psychrometer chamber assemblies were placed in a water bath inside an insulated box and allowed to equilibrate in the laboratory. Measurements were taken with a 12-channel digital psychrometer meter (85 series, JRD Merrill Specialty Equipment) in the psychrometric mode.

Stem water storage calculations

Branch sap flow rates were used to obtain whole-crown transpiration by first computing average values for all three to four branches and normalizing these with respect to the average daily maximum value. Normalized branch sap flow was then divided by the daily sum of the normalized 10 min averages divided by 6, then multiplied by the total daily sap flow measured at the base of the tree which was assumed to be equal to total daily transpiration. This procedure yielded estimated rates of crown transpiration on an hourly basis. Total diurnal stem water storage capacity was estimated by subtracting 10 min averages of basal sap flow from whole-crown sap flow when basal flow was less than whole-crown flow, summing the differences, then dividing by 6.


Sapwood area and total daily sap flow measured at the base of the trees were positively correlated with tree height (r = 0·86, P < 0·01; r = 0·88, P < 0·01, respectively), with the tallest tree, A. excelsum, having the largest sapwood area and sap flow (both measured at the base of the tree) and the smallest tree, C. longipes, having the smallest sapwood area and total sap flow (Table 1). Total daily sap flow and maximum transpiration rates (crown sap flow) were also positively correlated (r = 0·99, P < 0·01). For example, F. insipida had a daily total sap flow of 164 kg d–1 and a maximum sap flow rate of 17·2 kg h–1 while S. mombin had both lower daily total sap flow and maximum transpiration rates (80 kg d–1 and 10·5 kg h–1, respectively).

On days with similar environmental conditions, daily courses of basal and branch sap flow exhibited similar overall trends among all the trees (Fig. 1). Sap flow increased rapidly in the morning with increasing PPFD and VPD and decreased sharply in the afternoon for all the trees. However, the initial morning increase in basal sap flow lagged ≈ 0·1–1 h behind that of branch sap flow, depending on the individual (Fig. 1). The time lag between attainment of maximum sap flow rates in the upper branches and base of the stem tended to be even greater, ranging from 4 to 5 h in larger trees such as A. excelsum and F. insipida (Figs 1a & b), to about 1 h in L. seemannii and S. mombin (Figs 1c & d). In the smallest tree, an individual of C. longipes, maximum sap flow was observed almost simultaneously in the crown and at the stem base (Fig. 1e). The afternoon decrease in sap flow tended to occur earlier in branches than at the base of the stem, consistent with the morning trend. The rapid early morning increase in branch sap flow, followed by the increase in basal sap flow after a lag phase, suggested that stem water located between the base of the trunk and the upper branches was available for transpiration. Differences in the time courses of basal and branch sap flow were utilized to determine periods of withdrawal of water from and recharge of water to internal storage. Water flowing through terminal branches was considered to be derived mainly from internal storage when crown sap flow (the integration of sap flows for all branches in a tree) was greater than basal sap flow (Fig. 2). When the opposite pattern was observed, that is, when basal sap flow was greater than crown sap flow, the internal water stores were presumably being recharged (Fig. 2). In A. excelsum, for example, utilization of internal water occurred between 0800 and 1100 h, while the recharge of the internal water storage occurred primarily between 1300 and 1700 h (Fig. 2a). Basal sap flow in all trees typically returned to zero before 0600 h, suggesting that nocturnal recharge of sapwood water storage was essentially complete.

Figure 1.

. Daily courses of branch sap flow (dotted line) and basal sap flow (solid line) in (a) Anacardium excelsum, (b) Ficus insipida, (c) Luehea seemannii, (d) Spondias mombin, and (e) Cecropia longipes during representative clear days.

Figure 2.

. Difference between crown sap flow and basal sap flow for the same species on the same representative days as in Fig. 1. Positive values indicate time periods when water transpired was preferentially withdrawn from the stem water storage, and negative values indicate time periods when water from the soil was refilling stem storage.

The total diurnal water storage capacity differed between trees, ranging from 4 kg in C. longipes to 54 kg in A. excelsum. This represented a 67% increase, from 9 to 15% of the total daily water loss with increasing tree size (Table 1). All five individuals studied conformed to the same relationship between diurnal storage capacity and sapwood area (Fig. 3a). The magnitude of diurnal water storage appeared to increase linearly with increasing sapwood area with an increment of about 10 kg of stored water for every 0·1 m2 increase in basal sapwood area. All five trees also conformed to a common relationship between basal sap flow and sapwood area (Fig. 3b) The relationship appeared to be moderately asymptotic, with sap flow increasing about 125 kg d–1 for every 0·1 m2 increase in basal sapwood area in the three individuals with the smallest sapwood area and about 65 kg d–1 for every 0·1 m2 increase in sapwood area over the size range encompassed by the two largest trees.

Figure 3.

. Diurnal water storage (a) and total daily sap flow (b) in relation to sapwood area for the five study trees.

Diurnal water storage capacity increased exponentially with tree height, resulting in large increases in diurnal storage with small increases in tree height in larger trees (Fig. 4). For example, diurnal storage capacity increased only about 5 kg with an increase in tree height of 8 m from 17 to about 24 m, resulting in an increment of about 0·6 kg m–1 increase in height. The corresponding diurnal water storage increment over the range of height encompassed by the three largest trees was about 6 kg m–1, an order of magnitude greater.

Figure 4.

. Diurnal water storage in relation to tree height for the five study trees. Symbols as in Fig. 3.

Under fluctuating environmental conditions, water was alternately withdrawn from internal storage followed by periods of recharge during the day. For example, on 27 February 1996 passing clouds caused large fluctuations in PPFD and VPD (Fig. 5a). Following a typical early morning lag phase with respect to branch sap flow, basal sap flow in F. insipida increased rapidly at 0800 h and began to decrease at 1200 h, also lagging behind the decrease in branch sap flow (Fig. 5b). At about 1400 h branch sap flow increased again followed by an increase in basal sap flow. This resulted in two main periods of utilization and refill of internal water stores. The first period of utilization of stored water occurred between 0730 and 0930 h and the second period started at 1400 h and lasted for less than 1 h (Fig. 5c). The discharge and refill of the internal water storage was thus a dynamic process and fluctuated according to environmental conditions.

Figure 5.

. Diurnal courses of (a) photosynthetic photon flux density (PPFD, solid line) and vapour pressure difference between leaf and air (VPD, broken line), (b) basal sap flow (solid line) and branch sap flow (broken line), and (c) the difference between crown and basal sap flow in Ficus insipida.

The diurnal pattern of crown sap flow (a surrogate for transpiration) was influenced by the size of the internal water reservoir. Individuals having larger internal water storage capacity maintained maximum or near maximum transpiration rates for a longer period of time (Fig. 6). For example, crown sap flow remained near its maximum value for about 7 h in A. excelsum, a species with a diurnal water storage capacity of 54 kg (Fig. 6a), while the crown sap flow in L. seemannii, a species with a diurnal water storage capacity of 16 kg, remained near its maximum value for less than 4 h (Fig. 6c). The ratio of crown sap flow in the early afternoon and the maximum crown sap flow attained in the morning (E1300/Emax) was calculated as a measure of the maintenance of maximum crown sap flows during the day, with higher ratios indicating that maximum transpiration was maintained for a longer period of time. The E1300/Emax ratio increased asymptotically both with increasing diurnal water storage capacity and with increasing sapwood area (Fig. 7).

Figure 6.

. Average crown sap flow normalized with respect to maximum sap flow in (a) Anacardium excelsum, (b) Ficus insipida, (c) Luehea seemannii, (d) Spondias mombin, and (e) Cecropia longipes during two to five representative clear days.

Figure 7.

. The ratio of branch sap flow measured at 1300 h and maximum branch sap flow (E1300/Emax) as a function of (a) diurnal water storage and (b) sapwood area for the five study trees. Symbols as in Fig. 3.

The relationships between basal sap flow and evaporative demand (ASD) were associated with differences in tree size and therefore diurnal water storage capacity. When storage capacity was large, such as in A. excelsum, the response of basal sap flow to increasing ASD was linear (Fig. 8). When storage capacity was small, such as in S. mombin and C. longipes, the relationship between basal sap flow and ASD was clearly asymptotic. Minimum leaf water potentials tended to be more negative in the three largest trees but did not vary in a continuous manner with increasing tree height (Table 1). The soil-to-leaf hydraulic conductance, the ratio of crown sap flow per unit leaf area to the soil–leaf water potential gradient, was not correlated with tree size (Table 1). Variation in soil-to-leaf hydraulic conductance among species appeared to be associated with variation in the ratio of leaf area to sapwood area in terminal branches rather than tree height (Andrade et al. 1998).

Figure 8.

. Basal sap flow in relation to atmospheric saturation deficit (ASD) for the five study trees. Symbols as in Fig. 3.

After the initial depletion of the internal water reservoir in the morning, variations in basal and crown sap flow were more synchronized. This can be seen qualitatively in Fig. 1 and was more pronounced in the individuals with relatively small internal water storage capacity. For example, branch sap flow peaked shortly before 0900 h in S. mombin, then declined abruptly and fluctuated during the rest of the day (Fig. 1d), with the sudden drop after 0900 h occurring in spite of increasing VPD and PPFD. Basal sap flow, on the other hand, increased steadily from 0730 to 1030 h, but at a lower rate than branch sap flow, because water was preferentially being withdrawn from storage in the stem above the point of probe insertion (Fig. 2). Fluctuations in sap flow at the top and at the base of the tree appeared to be synchronized after about 1030 h, when stem water was no longer readily available.


Diurnal water storage capacity is defined in this study as the mass of water that can be withdrawn from the main trunk and branches during the day and replaced over a 24 h cycle. Other definitions of internal water storage capacity have focused on the total amount of water in a particular tissue compartment in relation to the total transpiring leaf area (Goldstein et al. 1984; Meinzer & Goldstein 1986). Plant size is often the major parameter determining water storage capacity. For example, Nobel & Jordan (1983) observed that the water storage capacity among leaves, stems and roots of three desert perennial plants was mainly due to difference in organ size. If the total amount of water that is withdrawn from the storage compartment and used in replenishing the water lost by transpiration is not known, then the water storage capacity can be defined as the amount of water that can be withdrawn for a given change in water potential of the storage compartment or tissue. This change in water content per unit change in water potential is generally referred to as the tissue's capacitance (Jarvis 1975; Tyree & Jarvis 1982).

The finding that individuals of all five diverse tree species studied conformed to a common relationship between diurnal storage capacity and sapwood area was unexpected. According to this relationship, basal sapwood area was a good predictor of the diurnal water storage capacity in the stems of the trees (Fig. 3a). However, estimates of diurnal water storage capacity were not entirely independent of measurements of sapwood area because the sap flow rates used to calculate diurnal storage were obtained by multiplying sap flux density by sapwood area. On the other hand, a plot of diurnal storage against independent measurements of tree height also yielded a common relationship between storage and tree size for all species (Fig. 4). Thus, the size of a tree rather than the species identity was the major determinant of its internal water storage capacity under the conditions of our study. The predominance of size over phylogeny in determining water storage characteristics was somewhat surprising given the relatively broad range of leaf phenology, successional stage, and crown architecture represented by the five study species. However, relationships between sapwood area and water storage capacity may differ in co-occurring species growing in other sites and/or in tree species with different functional traits (e.g. drought-deciduous) or when soil water is more limiting. When working at the leaf level, water storage, and therefore sapwood area, must be taken into account in comparing and interpreting variation in behaviour among individuals of different and even the same species. This finding raises questions concerning the interpretation of apparent interspecific differences in transpiration rates based on field measurements of single leaves because these rates may partially reflect variation in the stem water storage capacity of individual trees rather than species-specific characteristics of the leaves.

The diurnal patterns of exchange of water between the transpiration stream and storage compartments in the stem (e.g. Fig. 5c) suggest that the hydraulic connections between the transpiration stream and the intracellular and/or extracellular water of the sapwood were good in the five species studied. Nearly all woody plants contain tissues within or associated with the water transport pathway that could function in water storage. Living cells of the sapwood with elastic walls, which can undergo substantial changes in volume with relatively small changes in turgor, are well suited as intracellular water storage elements (Holbrook 1995). Extracellular water stores include water retained within intercellular spaces and the lumens of cavitated xylem elements (Tyree & Yang 1990). Capillary water and water released by cavitation may account for a certain fraction of the readily available extracellular water in the sapwood (Zimmermann 1983; Tyree & Yang 1990). In particular, capillary water retained by low surface tension in apoplastic spaces before water potential declines during periods of high transpiration, could represent a substantial fraction of the water withdrawn from storage in the stem (Holbrook 1995). Under conditions of severe drought, on the other hand, water released by cavitation may aid in survival by preventing the desiccation of fast growing tissues (Dixon et al. 1984; Tyree & Yang 1990).

Although sapwood has diffuse boundaries with adjacent tissues and may function both in storage and conduction, parenchymatous tissues such as the pith in stems of some woody plants, have clearly defined boundaries, and only serve, with few exceptions, as internal water reservoirs (Goldstein et al. 1984; Franco-Vizcaino et al. 1990; Holbrook & Sinclair 1992a,b). In particular, the well developed inner and outer parenchyma tissues of Fouqueria columnaris, a tree species endemic to arid and semiarid regions of Mexico, can supply the leaves with sufficient water for several weeks during the dry season (Nilsen et al. 1990; Franco-Vizcaino et al. 1990). Caulescent giant rosette species growing in tropical alpine environments have a distinct succulent pith in the centre of a woody stem thermally insulated by a thick layer of marcescent leaves (Goldstein et al. 1984; Meinzer & Goldstein 1986). The amount of water released from pith water storage into the transpiration stream of giant rosette plants growing at 4200 m elevation is adequate for avoiding leaf water stress in the early morning when soil water is frozen and root water uptake is impaired by low temperatures (Goldstein et al. 1984). These elastic parenchymatous tissues whose function appears to be mainly water storage have the disadvantage of being at some distance from the main vascular conduits. In order to serve effectively as a water reservoir in tall trees, these tissues have to be in relatively close contact and have good hydraulic connections with the sapwood. These functional constraints may have prevented elastic stem tissues from developing into more conspicuous water storage compartments in large trees.

With the onset of transpiration in the morning, all of the transpired water was initially derived from stem storage (Fig. 1). After about 1 h or less, sap flow at the base of the stem commenced and the transpiration stream contained a mixture of water derived from external and internal sources. The contribution of stem water storage to the total daily transpiration ranged from 9 to 15% in the five trees studied, and was within the range of values estimated for other tree species (e.g. Schulze et al. 1985; Tyree & Yang 1990; Loustau et al. 1996). Although the amount of water consumption form internal storage was relatively small compared to the total daily water uptake from the soil, trees with greater water storage capacity sustained maximum transpiration, and therefore kept their stomata open, for longer periods of time (Figs 6 & 7). In addition, basal sap flow increased linearly with evaporative demand in the tree with the largest diurnal water storage capacity and asymptotically in the trees with the smaller storage capacities (Fig. 8). Taken together, these results suggest that stomatal limitation of transpiration in the crowns of smaller trees was stronger than in larger trees. Water storage capacity, or other tree size-dependent characteristics thus clearly influenced the diurnal course of transpiration and the response of basal sap flow to evaporative demand. The extent to which these relationships would be maintained if sapwood water stores were to become progressively depleted during a soil drying cycle is not known.

In forest trees, increasing height may contribute to maximization of light interception and therefore carbon gain. Nevertheless, as the upper canopy is approached the concomitant increase in the length of the water transport pathway may imply a potential trade-off in terms of increasing resistance to water movement from the soil to the leaves. Substantial increases in hydraulic resistance with increasing tree height could lead to dysfunctional cavitation in the xylem, unless partial stomatal closure limits transpiration rates. On the other hand, internal water storage may partially compensate for the tendency of hydraulic resistance to increase with tree height thereby limiting the development of diurnal leaf water deficits with no associated cost in terms of stomatal limitation of carbon gain when soil moisture is adequate. Another alternative for limiting the magnitude of leaf water deficits with increasing height is a reduction in total leaf surface area in relation to sapwood area, which would enhance the tree's relative water transport efficiency because each remaining leaf would be supplied by a larger fraction of the conducting system.

It is highly unlikely that the exponential increase in diurnal water storage capacity with tree height (Fig. 4) reflected differences in tree hierarchy (suppressed versus emergent) within the canopy. The canopy of the secondary seasonal forest where our study was conducted is relatively open. The crown of the shortest study tree, C. longipes, was in a large gap and was not overshadowed by the crowns of larger trees. The remaining study trees were similarly exposed. The exponential increase in water storage capacity with tree height and the lack of dependence of leaf-specific hydraulic conductance on tree height (Table 1) therefore further suggest that water supply to the terminal branches did not become increasingly limited by axial hydraulic resistance as tree height increased.

It has been suggested that transpiration and net CO2 uptake, and therefore tree growth, are limited by increasing axial hydraulic resistance in old and tall trees (Ryan & Yoder 1997). Evidence for this hypothesis comes largely from studies carried out with conifers from temperate climates (e.g. Waring & Sylvester 1994; Mencuccini & Grace 1996). The apparent inconsistency between our results and those cited above may reflect intrinsic differences in hydraulic architecture and life history traits between temperate conifers and tropical canopy angiosperm trees. In temperate conifer species, the seasonal duration of sapwood production and functioning is more restricted than in tropical species, possibly resulting in smaller sapwood area and volume in temperate compared to tropical tree species of similar size. Furthermore, annual growth rates tend to be higher and the maximum height and life span of the individual trees substantially shorter in tropical canopy trees. Average life expectancy for canopy tree species on Barro Colorado Island, Panama, a somewhat wetter seasonal forest than our canopy study site is about 96 years (Putz & Milton 1982). The cause of death for tropical canopy trees may be more linked to catastrophic biomechanical failure than to hydraulic limitations on water supply to the upper crown. In the Barro Colorado Island forest, for example, 60% of the tree deaths were caused by snapped trunks, 17% by uprooting, and only 14% of the trees died standing (Putz & Milton 1982). Tree fall most frequently occurs during the middle of the wet season rather than during the dry season (Brokaw 1982). The validity of the hydraulic limitation hypothesis in tropical canopy trees needs to be tested with exposed trees representing a range of sizes and species. Observations should be made during the dry season, when stem water storage may be progressively drawn down and stomata eventually forced to close. In our study, measurements were made during a dry season period characterized by substantially higher than normal rainfall.


Overall, the results of this study indicate that there was substantial utilization of water stored in the stems of five canopy trees in a seasonal tropical forest. The stored water was used primarily in the morning to replace transpirational losses before water contained in the soil and lower stem reached the upper crown. Stem water storage may partially compensate for the tendency of hydraulic resistance to increase with the length of the water transport pathway and play an important role in regulating the water status of leaves of tall canopy trees under the large diurnal variations in evaporative demand that exist in seasonal tropical forest ecosystems. Tree size (sapwood area and volume) rather than species was the most important determinant of diurnal storage. Therefore, patterns of crown sap flow or transpiration observed in an individual tree at a given time may be size- rather than species-specific. Our results further suggest that when working at the leaf level, sapwood area (water storage capacity) must be taken into account in comparing and interpreting behaviour among individuals of different, and even the same species, because differences in stomatal regulation of leaf gas exchange may reflect differences in stem water storage capacity rather than intrinsic species traits. Additional research is necessary to fully understand the role of stem water storage in the regulation of water flux through trunks, branches and leaves of canopy trees. More measurements on individuals encompassing a range of sizes within species, as well as measurements on additional species, will allow broader conclusions about the physiological and ecological consequences of stem water storage to be drawn.


This research was supported by National Science Foundation grant IBN-94–19500. We are grateful to the Smithsonian Tropical Research Institute for its support, and to Jose Herrera and Marek Zdzikot for their skill and patience in operating the crane.