key: cord-009804-4lozmf0h
authors: Lion, Marryanna; Kosugi, Yoshiko; Takanashi, Satoru; Noguchi, Shoji; Itoh, Masayuki; Katsuyama, Masanori; Matsuo, Naoko; Shamsuddin, Siti‐Aisah
title: Evapotranspiration and water source of a tropical rainforest in peninsular Malaysia
date: 2017-10-19
journal: Hydrol Process
DOI: 10.1002/hyp.11360
sha: 
doc_id: 9804
cord_uid: 4lozmf0h

To evaluate water use and the supporting water source of a tropical rainforest, a 4‐year assessment of evapotranspiration (ET) was conducted in Pasoh Forest Reserve, a lowland dipterocarp forest in Peninsular Malaysia. The eddy covariance method and isotope signals of rain, plant, soil, and stream waters were used to determine forest water sources under different moisture conditions. Four sampling events were conducted to collect soil and plant twig samples in wet, moderate, dry, and very dry conditions for the identification of isotopic signals. Annual ET from 2012 to 2015 was quite stable with an average of 1,182 ± 26 mm, and a substantial daily ET was observed even during drought periods, although some decline was observed, corresponding with volumetric soil water content. During the wet period, water for ET was supplied from the surface soil layer between 0 and 0.5 m, whereas in the dry period, approximately 50% to 90% was supplied from the deeper soil layer below 0.5‐m depth, originating from water precipitated several months previously at this forest. Isotope signatures demonstrated that the water sources of the plants, soil, and stream were all different. Water in plants was often different from soil water, probably because plant water came from a different source than water that was strongly bound to the soil particles. Plants showed no preference for soil depth with their size, whereas the existence of storage water in the xylem was suggested. The evapotranspiration at this forest is balanced and maintained using most of the available water sources except for a proportion of rapid response run‐off.

The response of tropical rainforests to environmental disturbance is of critical concern given the importance of their sustainability and potential to mitigate climate change. In these forests, changes in the amount and/or pattern of rainfall are perhaps the most important environmental disturbance. An understanding of fluctuations in evapotranspiration (ET) under different moisture conditions is necessary to determine how tropical rainforests will respond to climate change.

Research to date has demonstrated that tropical rainforests maintain ET even during the dry period. In Pasoh Forest Reserve (FR), which is located in a dry zone of Peninsular Malaysia and receives the lowest yearly rainfall amount among adjacent south-eastern tropical rainforests, relatively stable ET, which includes transpiration, interception evaporation, and soil evaporation, was observed even during the driest period, based on 7 years of continuous eddy covariance (EC) measurement (Kosugi, Takanashi, Tani, et al., 2012) . Consequently, stable annual ET rates (1,287 ± 52 mm) were obtained despite the relatively small annual rainfall amount (1,805 ± 280 mm, from 1995 to 2015) (Kosugi, Takanashi, Tani, et al., 2012) compared to other Southeast Asian tropical rainforests (e.g., Kume et al., 2011; Vernimmen, Bruijnzeel, Romdoni, & Proctor, 2007; Noguchi, Abdul Rahim, & Tani, 2003 ; Kumagai et al., 2005) . No obvious decline in monthly ET The English in this document has been checked by at least two professional editors, both native speakers of English. For a certificate, please see http://www. textcheck.com/certificate/bLyfRU variability was detected even during the driest month, although the amount of rainfall was much lower than ET (Kosugi, Takanashi, Tani, et al., 2012) .

Several other EC flux studies conducted in Amazonian and Southeast Asian tropical rainforests have also consistently demonstrated this characteristic (e.g., Costa et al., 2010; Da Rocha et al., 2004; Kume et al., 2011; Malhi et al., 2002) . In the Amazonian rainforests, ET did not drastically decrease and sometimes increased during the dry season (e.g., Costa et al., 2010; Da Rocha et al., 2004) . The Amazonian rainforest has distinct dry and wet periods, which is not the case in equatorial Southeast Asian rainforests, although dry and wet periods do exist as part of seasonal fluctuations, with considerable variability between years (Kosugi et al., 2008; Tani et al., 2003) . The stability of ET in dry periods and in the dry season seen in tropical rainforests should be supported by stable water sources in the soil throughout the seasons.

Hydrogen and oxygen isotopes (δ 18 O and δ 2 H) provide a powerful tool for determining plant water resources under a number of environmental conditions (Brooks, Barnard, Coulombe, & McDonnell, 2010; Dawson & Ehleringer, 1991; Eggermeyer et al., 2009; Evaristo, Jasechko, & Mcdonnell, 2015; West et al., 2012; Yang, Wen, & Sun, 2015) . Using isotope indices, we can investigate whether soil water, streamflow, and plant transpiration are all sourced and mediated by the same well-mixed water reservoir originating in the soil (Evaristo et al., 2015) . The depth of water uptake can also be estimated by measuring the isotopic composition of xylem water and soil water at different depths (Ehleringer & Dawson, 1992; Yang et al., 2015) . Many previous studies utilize only one of the dual stable isotopes of δ 18 O (Nie et al., 2011; Querejeta, Estrada-Medina, Allen, & Jimenez-Osornio, 2007) or δ 2 H (Filella & Penuelas, 2003; Jackson, Cavelier, Goldstein, Meinzer, & Holbrook, 1995) . However, the dual-isotope method (e.g., Cramer, Thorburn, & Fraser, 1999; Eggermeyer et al., 2009; Evaristo et al., 2015; S. G. Li et al., 2006; Orlowski, Frede, Brüggemann, & Breuer, 2013; Rossatto, Silva, Villalobos-Vega, Sternberg, & Franco, 2012; Wang, Song, Han, Zhang, & Liu, 2010; West, Hultine, Burtch, & Ehleringer, 2007) is a powerful tool for investigating the water source for ET. The use of a dual-isotope approach could systematically help in the separation of distinct water pools in the ecosystem. These pools usually include water used by trees that does not contribute to streamflow or mobile water that is unrelated to the water used by trees (i.e., groundwater, streamflow, infiltration, and hill slope run-off; McDonnell, 2014) . This approach can improve our understanding of the ecohydrological processes controlling water flows in soil-plantatmosphere continuums (Berry et al., 2016) .

Our study of ET and water sources in the Pasoh FR comprised three objectives: (a) measure and calculate ET using the EC method over a 4-year period (2012-2015); (b) determine spatial and temporal patterns of water uptake and provenance, using water budget methods combining ET, precipitation, and soil moisture data; and (c) determine the provenance of water that is transpired at different times of the year by assessing the stable isotope signatures of water in precipitation, soils, plants, and streams. This information is expected to aid understanding of the likely impact of climate change on water demand and supply in tropical rainforests.

The study was conducted in a lowland dipterocarp forest within the 6 ha of Pasoh FR (Figure 1 ) located at 2°58′N, 102°18′E at approximately 75 to 150 m above sea level. The soil in this area belongs to the local Durian series, which is classified as an ultisol with a yellowish silt-clay layer (40-80 cm thick), overlaying a blocky indurated lateritic horizon (30-40 cm thick) on top of mottled white clay that overlays weathered shale down to a depth of 130-150 cm (Leigh, 1982) . The A horizon was thin (0-2 cm) and consisted of brown silty loam, whereas deeper soils were bright yellowish or reddish brown and heavier (light to heavy clay; Yoda, 1978) . Lateritic gravel is abundant below 30 cm of depth and increases with depth Yoda, 1978) . The soil particle size distribution is characterized by low sand and high silt contents (Yamashita et al., 2003) . The Durian soil series at 50 cm has 1.50 g cm −3 bulk density as well as 39.6% moisture content at 0.33 bar and 8.3% between 0.33 and 15 bars. At the deeper soil layer (100 cm), bulk density was 1.41 g cm −3 , and moisture content was 46.2% (0.33 bar) and 22. 5% (0.33-15 bar, Universiti Pertanian Malaysia, 1979) . The underlying rocks at Pasoh FR include shales and sandstones from the Upper Triassic age (Gobbett, 1972; Peh, 1978) . The textural details of this soil based on Peh (1978) are listed in Table 1 . The maximum depth of tap roots was about 4 m (Niiyama et al., 2010) ; most of the fine roots were found at the A horizon (Amir Husni, 1989) . Most fine roots that have a diameter of 1-2 mm were found at 0-4cm soil depth, whereas fine roots with a diameter of 3-5 mm were in abundance at 12-16-cm soil depth (based on Figure 7 in Yamashita et al., 2003) . Fine roots play an important role in absorbing water and nutrients and in dry matter production. The total fine root biomass down to 2-m depth was estimated at an average of 13.3 Mg ha −1 using a pipe model approach and 16.4 Mg ha −1 using a pit sampling method (Niiyama et al., 2010) .

The canopy height ranges between 30 and 40 m with emergent trees of~45 m. Rainfall distribution in Pasoh FR is of short duration and high intensity, with a mean annual rainfall of 1,805 mm and mean annual air temperature of 25.4°C (1997) (1998) (1999) (2000) (2001) (2002) (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (Noguchi et al., 2003 (Noguchi et al., , 2016 . Streams in this area are typically shallow with relatively low baseflow (Leigh, 1978) .

Direct run-off in response to rainfall is dominant, and the streams sometimes dry up during drier periods. Continuous EC flux observations have been conducted at the microclimate observation tower since September 2002 (Kosugi, Takanashi, Tani, et al., 2012; Kosugi et al., 2008; Takanashi et al., 2010) and showed very stable average monthly diurnal ET courses. Manokaran (1979) reported an annual interception value of 519 mm (21.8%) for Pasoh FR. Subsequently, Tani et al. (2003) recorded an annual interception for Pasoh FR from 1999 to 2000 of 381.3 mm (16.9%). Throughfall ranged between 79% and 94% in this forest (Konishi et al., 2006) . In terms of run-off generation, a substantial amount of overland flow was measured using the trap method, indicating a close relationship with the amount of rainfall (Leigh, 1978) . Data were sampled at 10 Hz and sent to a data logger (CR-5000 or CR-1000, Campbell Scientific, Logan, UT). Momentum, sensible heat (H, W m −2 ), and latent heat (ƛE, W m −2 ) fluxes were calculated as 30min averages. Double rotation (McMillen, 1988) was applied, assuming a zero mean vertical wind. The Webb-Pearman-Leuning (Webb, Pearman, & Leuning, 1980) correction for the effect of air density fluctuations was also applied, and linear trends in temperature and water vapour were retained. Our EC measurements of ET included transpiration, interception evaporation, and soil evaporation. We discarded all latent heat flux data recorded during and after rainfall; however, this gap was filled using available energy and sensible heat flux data measured using a three-dimensional ultrasonic anemometer, which can collect data during rainfall. A detailed description of these methods and instruments is provided by Kosugi and colleagues (Kosugi, Takanashi, Tani, et al., 2012; Kosugi et al., 2008) . The energy budget correction was executed to overcome the energy imbalance problem described by Takanashi et al. (2010) in estimating ET. Both sensible and latent heat fluxes are corrected using the Bowen ratio to produce zero energy imbalance (Kosugi, Takanashi, Tani, et al., 2012) . Gaps were further filled using a second-order polynomial relationship between 30min λEs after the energy budget correction method, and the available The energy released by changes in the air, vapour, and trunk storage was estimated using temperature and vapour pressure differences at the top of the tower based on Ohtani et al. (1997) . Air temperature and humidity at a height of 52 m were observed using an HMP45A or HMP45C instrument (Vaisala, Vantaa, Finland). An Assmann psychrometer (SY-3D, Yoshino Keiki, Japan) was used to calibrate the Vaisala sensors periodically. The vapour pressure deficit (VPD) of the air was calculated from these data. Tipping-bucket rain gauges (Ota Keiki 34-T, Japan) were used to measure rainfall at the top of the 52-m flux tower and at an observatory located 430 m away from the tower. These data were compared and corrected with the rainfall measured with a storage gauge at the observatory. The volumetric soil water content (VSWC) was measured using time domain reflectometry sensors (CS615 or CS616, Campbell Scientific) at depths of 0.1, 0.2, and 0.3 m at three points around the tower logged at the 30-min intervals (Noguchi et al., 2016) . This layer contains the majority of the fine roots. The mean rooting depth in a mixed dipterocarp forest has been identified as 2.35 m (Baillie & Mamit, 1983) . We used the daily average value of these nine sensors as a reference VSWC for the surface layer between 0 and 0.5 m. The time domain reflectometry sensors were calibrated using the standard procedure for calibrating capacitance sensors outlined by Starr and Paltineanu (2002) , which consists of nine steps (Noguchi et al., 2016) . Solar noon in this site peaks around 01:00 p.m. local time (Kosugi, Takanashi, Tani, et al., 2012) . The antecedent precipitation index (API60) was tested and found to have a significant relationship with the VSWC (Noguchi et al., 2016) and was therefore used as a wetness index in this study area. The API60 is defined as ∑ 60 i¼1 P i =i, where P i is the daily precipitation (mm) and i is the number of preceding days (Kosugi et al., 2007). 2.3 | Isotope signatures of water in precipitation, stream, plants, and soil A storage-type rain gauge with a double-layered small-mouth inner glass bottle was installed at the observatory station to collect rainwater samples and was buried in the ground to prevent heating and evaporation. The storage gauge has a collar and is surrounded with a sponge to prevent splash-in from surrounding soil during rainfall. Rainwater samples for isotope analysis were collected daily at 8:00-9:00 a.m. ; however, these samples were not taken during the rainfall.

We collected eight samples when there had been no rainfall for more than a week. Samples were filtered and transferred into two 10-ml polyethylene terephthalate bottles for stable isotope analysis. There was no rain over the course of these two sampling days. Soil samples (n = 3) were obtained using a hand auger at each depth, placed in 30-ml vials, sealed, and placed in a cool box to prevent evaporation.

All plant and soil samples were frozen in a refrigerator (−15 to −20°C)

in the laboratory before water extraction. Soil sampling was undertaken with care to prevent soil mixing between preceding layers and the current soil layer. Auger holes were dredged several times before storage to ensure no material from the preceding layer was included in the current sample. Tiny roots, stones, and leaf litter were separated from samples to prevent contamination. Water extraction was conducted using a cryogenic vacuum distillation system (West, Patrickson, & Ehleringer, 2006) . Cryogenic vacuum extraction is the most widely utilized method for plant and soil water extraction (Ingraham & Shadel, 1992; Koeniger, Marshall, Link, & Mulch, 2011; Orlowski et al., 2013; Vendramini & da S. L. Sternberg, 2007; West et al., 2006) . We treated water extracted from plants with granulated activated charcoal to adsorb organic compounds that may influence the isotope contents of the samples (West et al., 2006) . The samples treated with granulated activated charcoal were kept for 1 week before being transferred into measurement vials.

All water samples, including precipitation, stream, plants, and soil, were filtered using a phobic polytetrafluoroethylene filter (with 13mm diameter and 0.45-μm pore size) and transferred into 2-ml vials (C5000-54G) in the laboratory for isotope analysis. A cavity ringdown spectrometer (L2120-i, Picarro, CA) was used to analyse the isotope composition of the samples; this device had a specific analytical precision of 0.06‰ for δ 18 O and 0.11‰ or less for δ 2 H (Katsuyama, 2014) . The delta (δ) The ET reaches a ceiling at high VPD, whereas at low VSWC during drier periods, a small decrease in ET was observed with high VPD, indicating stomatal closure ( Figure 3b ).

Comparing ET with water budget in the soil provides some insight into the source of water for ET (Figures 4 and 5) . water was required in Pasoh FR to accommodate ET demand. A substantial amount of overland flow at the slope of this site was previously observed (Leigh, 1978; Peh, 1978) . However, some portion of such overland flow is assumed to infiltrate before reaching the stream. If this flow were taken into account, considerable water supply to the soil would be expected. Nevertheless, the water supply would still be less than indicated by the antecedent precipitation values shown in this figure.

Figure 5 examined the comparison between ET demand and water budget in the surface soil layer to evaluate the source depth of water for ET. The ET demand was calculated as the cumulative budget of rainfall and ET. If the budget was above 0, then the rainfall portion was not included. The soil water storage differential in soils at depths of 0-0.5 m was calculated using the VSWC, assuming uniform soil water content between 0 and 0.5 m. If both decline slopes are similar, this indicates that most water was supplied from this surface layer, Three rough groupings could be identified in the plots for plant water (Table 3) 

Pasoh FR has typical monsoon-type rainfall variations (Figure 2a,b) . It has significant interannual variations, with a bimodal pattern consisting of two distinct peaks; during March to May and October to December (Marryanna et al., 2017) . Using long-term average climate records, this area falls under the Af (tropical rainforest climate) in the Köppen-Geiger climate classification scheme and is located within an equatorial region that hosts a fully humid climate (all months of the year are warm). This is the synoptic climate of Southeast Asia that is Antecedent rainfall is also plotted to identify water sources Southwest China, showed a smaller value (1,029 ± 29 mm, Z. Li et al., 2010 Costa et al., 2010; Da Rocha et al., 2004; Kume et al., 2011; Malhi et al., 2002) . Nepstad et al. (1994) reported that some rainforests are able to extract soil water even in the dry season because of deep rooting depths. In Pasoh FR, the rooting depth was concentrated in the A horizon (Amir Husni, 1989) but extended down to 4 m (Niiyama et al., 2010) . On the basis of energy imbalance filtering data, ET was stable with a slight increase in the dry season in central Amazonian rainforest sites, whereas in drier semideciduous sites, it showed a slight decrease during the dry season . This site-dependent difference was caused by differences in the increasing radiation energy and the stomatal behaviour during the dry season. ET is mainly determined by atmospheric evaporative demand (e.g., Carswell et al., 2002; Da Rocha et al., 2004; Z. Li et al., 2010) and stomatal control (e.g., Cunningham, 2004) , the latter of which is influenced by the factors related to water supply, such as available water in the soil (e.g., Burgess, Adam, Turner, & Ong, 1998; Rafael, Todd, Burgess, & Nepstad, 2005) , rooting depth (e.g., Z. Li et al., 2010; Nepstad et al., 1994) , and upward flow in the soil (e.g., Da Rocha et al., 2004; Z. Li et al., 2010) . In Malaysian tropical rainforests located near the equator with indistinct dry periods, ET was relatively constant (Kumagai et al., 2005; Kume et al., 2011; Tani et al., 2003) . The variability of daily ET at Pasoh was dependent on available energy and VPD but was also moderately influenced by soil water content (Figure 3a ,b, Kosugi, Takanashi, Tani, et al., 2012) . (2010) could be the driving factor. The drought-tolerant tree and deep root system (Hodnett, Tomasella, Marques, De, & Oyama, 1996; Nepstad et al., 1994) could also be the underlying factor. ET decreased in the rainy season because the available energy and temperature decreased. It has been reported (Kosugi, Takanashi, Tani, et al., 2012 ) that this forest used on average (±standard deviation)

1,287 ± 52 mm year −1 for ET during 2003 and 2009. These values were slightly higher compared to those found in this study during 2012 and 2015 (1,182 ± 26 mm). This is probably because of the drier conditions during these 4 years.

Soil water content is a major control of hydrological processes such as ET, the precipitation run-off response, energy transfer, and a climate predictor (e.g., Betts, Ball, Beljaars, Miller, & Viterbo, 1996) . The soil water budget is a useful method for estimating total water loss from the soil caused by transpiration and soil evaporation (Wilson, Hanson, Mulholland, Baldocchi, & Wullschleger, 2001) once soil water content has dropped below field capacity. The other benefit of the soil water budget is that it can provide insight into the relative contribution of various rooting depths to total transpiration sources (e.g., Teskey & Sheriff, 1996) . This forest needs at least four months of water storage to accommodate the ET demand during extended dry periods ( Figure 4) . The typical clay soil in Pasoh FR would delay water infiltration into the soil during rainfall and therefore contribute to overland flow. This means that the required water storage period should be longer than the analysis shown in Figure 4 . During dry periods, soil may experience excessive drying, and plants therefore need to use the previous water storage that is more strongly bound to soil particles.

The slope of ET demand (Figure 5b) is an important indicator of the water change in the soil. When water supply from a specific soil layer is less than the ET demand, the forest may obtain supplies from other sources. The analysis in Figure 5 shows that in Pasoh FR, the forest mainly acquired water from the surface layer; however, during dry periods, insufficient water was available and water was acquired from deeper layers through the rooting system. The soil moisture content for soils at a depth of 0.05 m in this forest was reported to have an average field capacity of 34.6%, a permanent wilting point of 21.9%, and a saturation point of 46.9% (Peh, 1978) . In dry periods, the in Pasoh FR was about 4 m, and it was noted that fine roots were still found at 4-m depth (Niiyama et al., 2010) . This finding supports our result.

Rainwater is partitioned into three components: (a) interception loss, (b) infiltration excess overland flow, and (c) water infiltration to soil.

Infiltrated water is divided between percolating water held at tension below field capacity (drainable pore space), plant-available water held at tension between field capacity and permanent wilting point (water available for plant ET), and immobile residual water not participating in the hydrological cycle. Our results (Figures 6 and 7) strongly suggest that water sources for stream, plants, and soil at this forest, corresponding to infiltration excess overland flow + water infiltration to soil, plant-available water held at tension between field capacity and permanent wilting point, and immobile residual water not participating in the hydrological cycle, respectively, are separated.

Large fluctuations in isotope values in stream water corresponding with rainwater ( Figure 6a ) indicate that the portion of new water runoff is large in this watershed. Streams lose water flow during severe drought periods at this site. Due to the rainfall characteristics in Pasoh FR, which is intense and of short time duration (Noguchi et al., 2003) , and the clay soils with low permeability, overland flow is often observed at the slope (Leigh, 1978; Peh, 1978) . Our results are consistent with this. Generally (e.g., Clark & Fritz, 1997) , the isotope values in stream water will lie close to the long-term weighted mean value for rainwater because stored, well-mixed groundwater maintains the streamflow. We need more intensive data of stream water to check this consistency and to obtain more precise partitioning of run-off pathway in this site.

Soil and plant water showed different characteristics from stream water and, additionally, differed from each other (Figures 6 and 7) .

Only in the very dry period did plant and soil isotope values become closer, because plants used water from the same reservoir in the soil.

Clay soil typically has higher saturated and residual soil water content in the water retention curve, and hence a substantial portion of the water is strongly bound to soil particles and is not available to plants.

This most likely explains the different water sources between soils and plants at the study site. Interactions between the matrix and the gravitational potential lead to pores, with the largest diameter filling first and pores with the smallest diameter draining last and containing immobilised water compared to the larger pores (Brooks et al., 2010) .

Evaporation from soil decreases rapidly with depth, and thus, plant roots are the primary cause of soil drying to below field capacity. Additionally, water with the longest residence times in soil is more likely to be removed by plants and not delivered to the stream during dry seasons (Brooks et al., 2010) . This situation was also observed in our study during the driest sampling period (June 2013 and March 2014, Figure 7 a,b), when the isotope compositions of plant waters were similar to soil water and deviated from rainwater. This could indicate that plants are using water with the longest residence time and bound tightly to soil particles to supply their ET demand during dry periods. We did not test how tightly the water was bound to the soil, and this could be a significant topic for future study. However, it can at least be inferred from Figure 2 that the VSWC of soils from the study site was approximately 0.28 m 3 m −3 in the driest period; thus, this value should be close to the residual soil water content. This means that the soil contained more than 20% water that was very tightly bound to soil particles, and thus inaccessible to plants. This amount of water would be included in the soil even in the wettest period, and we can conclude that this is one of the main reasons for the isotopic difference between water from soil and plants. The two only became similar in the very dry period.

Both the deviation in isotope values for plant water to the right side of the LMWL and the similarity in the range of fluctuation in the isotope values of plant water and soil water at the surface soil layer suggest that a substantial part of plant water was supplied from the surface soil layer. However, it should also be noted that the isotope values of plant water could not be explained entirely by surface soil water isotope values; this suggests plant water uptake also includes water in deeper soil layers. This analysis is consistent with the results of the water budget analysis shown in Figure 5 .

It has been assumed that plants take up water according to their height (Kim, Park, & Hwang, 2014) and rooting depth (Meinzer, James, & Goldstein, 2004) . Water uptake and transport are associated with a hydraulic flow process that is controlled by resistance and hydraulic gradients (Honert, 1948) . The overall resistance is determined by soil water potential, conducting vessels, transpiration rate, plant height, and gravity (Kim et al., 2014) ; however, the results from this study do not necessarily reflect this trend. The isotope signals from trees of different height and species did not show any clear tendency of water uptake depth during the study period. From these results, we could not distinguish whether shorter or smaller trees use water from the shallow soil layer, whereas taller trees take water from the deeper soil layer. Water uptake by co-occurring woody species showed that some species only take water from deep or shallow soil, whereas others use both layers (Eggermeyer et al., 2009; West et al., 2007) . In tropical and subtropical forests, plants typically have shallow roots, meeting their water needs from the surface soil layer due to the plentiful rainfall (Schenk & Jackson, 2005) and soil properties would also be associated with variability in isotopic values (Orlowski, Pratt, & McDonnell, 2016) . The comparison of different extraction techniques for different soil types (sandy and clay soils) identified that differences between the methods were small in sandy soil but large in clay soils (Figueroa-Johnson, Tindall, & Friedel, 2007; Orlowski et al., 2016) . Despite this, there are still many factors that may influence isotopic unfractionated water, and as a result, this issue remains poorly understood and there is a need for further studies (Munksgaard, Cheesman, Wurster, Cernusak, & Bird, 2014; Orlowski et al., 2016) . The centrifugation method to extract soil water, which can collect only capillary water with a certain limit of capillary pressure, could be used for this purpose.

Our study also found some classifications for plants in terms of water isotope characteristics. Group 1 showed isotope values close to that of soil water. This group consisted of an emergent tree (D. sublamellatus), two forest floor trees (M. lowii and B. parviflora), and a middle-sized tree (S. rugosum). Group 2 showed more negative isotope values that were out of the range of soil water, but still within the range of the antecedent rainwaters. All three trees in Group 2 were middle-sized trees (X. stipitatum, H. dictyoneurum, and D. malaccensis).

The Group 3 tree (P. caput-medusae) was also middle sized and showed very stable and less negative isotope values during the whole observation period; however, these values differed from those of the deepest layer soil water. Some Group 2 (X. stipitatum) and Group 3 (P. caputmedusae) trees have small stomatal conductance and maximum photosynthesis (Kosugi, Takanashi, Yokoyama, et al., 2012) . These may be protective mechanisms allowing trees to retain more water in their xylem to prevent water loss, and the portion of storage water may be large. As the portion of newly absorbed water in xylem becomes smaller, the characteristics of isotope values would shift from that of Group 1 to Group 2 and finally to Group 3. Group 3 may store more water inside the xylem, rendering these plants more drought resistant.

All trees in Groups 2 and 3 are medium-sized trees, which may indicate the origin of these characteristics. We found smaller stomatal conductance and photosynthesis in middle-sized trees at this site compared with emergent trees (Kosugi, Takanashi, Yokoyama, et al., 2012) . These middle-sized trees might suffer from water stress and operate in a highly protective manner (Kosugi, Takanashi, Matsuo, & Abdul Rahim, 2009 ). The utilization of released water from stored compartments near the canopy may play an important role in mitigating water stress in canopy leaves, thereby maintaining stomatal opening for photosynthesis. Some previous studies have also shown that increased withdrawal of internally stored water occurs during water deficits (e.g., Scholz et al., 2007) , whereas others have shown that stem-stored water may be used for buffering the daily water deficit even when soil water is abundant (Goldstein et al., 1998; Holbrook, 1995) .

Pasoh FR used most of the available water to maintain ET particularly during dry periods in which there was no decrease of ET ( Figure 2d ).

During dry periods, that is, when surface soil water was not available, the forest's root system was able to access water from greater soil depths to maintain ET (Nepstad et al., 1994) corresponding to the existence of a deep unsaturated soil layer in this forest. Our sampling revealed no groundwater within 3 m of the surface, including during the rainy season. Streams in this area are typically shallow with relatively low baseflow (Leigh, 1978) . Direct run-off in response to rainfall is dominant, and the streams sometimes dry up during drier periods.

This forest is mainly characterized by clay soil (Yoda, 1978) , with low sand and high silt soil particle distributions (Yamashita et al., 2003) .

The physical traits of the clay soil likely influenced the water storage capacity of the watershed available for ET. Streamflow characteristics are typically determined by soil physical traits, topography and geography. In our study area, the uptake of almost all-available water by vegetation also influenced streamflow characteristics. In many tropical rainforests with rainfall amounts of~2,000 mm and having occasional dry period, the amount of run-off is not large, as illustrated by several reviews of hydrology in the humid tropical region (e.g., Wohl et al., 2012) . Rather, run-off is typically low and occasionally ephemeral during dry periods due to the large ET demands of rainforest ecosystems.

Overall, although ET during the observation years was lower than in previous studies, a stable pattern was observed. At least four months of water storage is needed to accommodate ET needs during a very dry season. Spatially, plants in Pasoh FR typically obtained their water supply from the surface soil layer (0-0.5 m) and sourced further water from deeper layers during the dry period. Deep rooting and stomatal control are two well-known mechanisms that allow plants to cope with periods of high atmospheric demand and low water availability. Isotope analysis results show that plants, soil, and stream have different sources of water in this forest. Similarity in rainwater and stream water indicates the dominance of new water run-off. There are occasional isotopic differences between plants and soil water, probably because water source for plants is different from water strongly bounded to the soil. The source of water for this forest does not have a distinct pattern corresponding to soil depth and tree height. The results also suggest the existence and use of water storage in tree xylem. ET at Pasoh FR is balanced and maintained using most of the available water sources except for a proportion of rapid response run-off. This study demonstrates that tropical rainforests show a degree of water stress;

under climate change, precipitation amounts and/or pattern are expected to change, which potentially cause not only ET but also shifts in vegetation patterns in the tropical zone.

We would like to acknowledge the contribution of a RONPAKU 

Site fertility and carrying capacity in two Malaysian tropical forest reserves

Observations on rooting in mixed dipterocarp forest

The two water world's hypothesis: Addressing multiple working hypotheses and proposing a way forward

The land surface-atmosphere interaction: A review based on observational and global modelling perspectives

Ecohydrologic separation of water between trees and streams in a Mediterranean climate

The redistribution of soil water by tree root systems

Seasonality in CO 2 and H 2 O flux at an eastern Amazonian rainforest

Environmental isotopes in hydrogeology

Atmospheric versus vegetation controls of Amazonian tropical rain forest evapotranspiration: Are the wet and seasonally dry rain forest any different

Transpiration and ground water uptake from farm forest plots of Casuarina glauca and Eucalyptus camaldulensis in saline areas of southeast Queensland

Agriculture Water Management

Stomatal sensitivity to vapour pressure deficit of temperate and tropical evergreen rainforest trees of Australia

Effect of 7 years of experimental drought on vegetation dynamics and biomass storage of an eastern Amazonian rainforest

Seasonality of water and heat fluxes over a tropical forest in eastern Amazonia

Streamside trees that do not use stream water: Evidence from hydrogen isotope ratios

Seasonal changes in depth of water uptake for encroaching trees Juniperus virginiana and Pinus ponderosa and two dominant C 4 grasses in a semiarid grassland

Water uptake by plants: Perspectives from stable isotope composition

Global separation of plant transpiration from groundwater and streamflow

A comparison of 18 Oδ composition of water extracted from suction lysimeters, centrifugation, and azeotropic distillation

Partitioning of water and nitrogen in cooccurring Mediterranean woody shrub species of different evolutionary history

Geology map of Malay Peninsula

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

Evapotranspiration of tropical peat swamp forest

Deep soil water uptake by forest and pasture in central Amazonia: Predictions from long-term daily rainfall data using a simple water balance model

Stem water storage

Water transport in plants as a catenary process

A comparison of the toluene distillation and vacuum/heat methods for extracting soil water for stable isotopic analysis

Partitioning of water resources among plants of a lowland tropical forest

Remarks of water isotope measurement of small samples with analyzer based on wavelength-scanned cavity ring-down spectroscopy (WS-CRDS)

Investigating water transport through the xylem network in vascular plants

An inexpensive, fast, and reliable method for vacuum extraction of soil and plant water for stable isotope analyses by mass spectrometry. Rapid Communications in Mass Spectrometry

Characteristics of spatial distribution of throughfall in a lowland tropical rainforest

Spatial and temporal variation in soil respiration in a Southeast Asian tropical rainforest

Midday depression of leaf CO 2 exchange within the crown of Dipterocarpus sublamellatus in a lowland dipterocarp forest in Peninsular Malaysia

Effect of inter-annual climate variability on evapotranspiration and canopy CO 2 exchange of a tropical rainforest in Peninsular Malaysia

CO 2 exchange of a tropical rainforest in Peninsular Malaysia

Vertical variation in leaf gas exchange parameters for a Southeast Asian tropical rainforest in Peninsular Malaysia

Annual water balance and seasonality of evapotranspiration in a Bornean tropical rainforest. Agriculture and Forest Meteorology

Ten-year evapotranspiration estimates in a Bornean tropical rainforest

Surface wash; experimental techniques and some preliminary results

Sediment transport by surface wash and throughflow at the Pasoh Forest Reserve

Seasonal variation in oxygen isotope composition of water for a montane larch forest in Mongolia

Evapotranspiration of a tropical rain forest in Xishuangbanna, Southwest China

Energy and water dynamics of a central Amazonian rain forest

Stemflow, throughfall and rainfall interception in a lowland tropical rain forest in Peninsular Malaysia. The Malaysian Forester

Temporal variation in stable isotopes in precipitation related with rainfall pattern in a tropical rainforest in peninsular Malaysia

The two water worlds hypothesis: Ecohydrological separation of water between streams and trees? WIREs Water

An eddy correlation technique with extended applicability to non-simple terrain

Dynamics of transpiration, sap flow and use of stored water in tropical forest canopy trees

Microwave extraction-isotope ratio infrared spectroscopy (MEIRIS): A novel technique for rapid extraction and in-line analysis of δ 18 O and δ 2 H values of water in plants, soils and insects

The role of deep roots in the hydrological and carbon cycles of Amazonian forests and pastures

Seasonal water use patterns of woody species growing on the continuous dolostone outcrops and nearby thin soils in subtropical China

Estimation of root biomass based on excavation of individual root systems in a primary dipterocarp forest in Pasoh Forest Reserve, Peninsular Malaysia

Rainfall characteristics of tropical rainforest at Pasoh Forest Reserve

Pasoh: Ecology of a lowland rainforest in southeast Asia

Long term variation in soil moisture in the Pasoh Forest Reserve, a lowland tropical rain forest in Malaysia

Energy and CO 2 fluxes above a tropical rain forest in Peninsular Malaysia

Validation and application of a cryogenic vacuum extraction system for soil and plant water extraction for isotope analysis

Intercomparison of soil water extraction methods for stable isotope analysis. Hydrological Processes

Rates of sediment transport by surface wash in three forested areas of Peninsular Malaysia

Water source partitioning among tree growing on shallow karst soils in a seasonally dry tropical climate

Hydraulic redistribution in three Amazonian trees

Depth of water uptake in woody plants relates to groundwater level and vegetation structure along a topographic gradient in a neotropical savanna

Mapping the global distribution of deep roots in relation to climate and soil characteristics

Biophysical properties and functional significance of stem water storage tissue in neo-tropical savanna trees

Methods for measurement of soil water content: Capacitance devices

Water and heat fluxes above a lowland dipterocarp forest in Peninsular Malaysia

A review of evapotranspiration estimates from tropical forests in Thailand and adjacent regions

Characteristics of energy exchange and surface conductance of a tropical rain forest in Peninsular Malaysia

Pasoh: Ecology of a lowland rain forest in Southeast Asia

Water use by Pinus radiata trees in a plantation

Soil physic annual report

A faster plant stems water extraction method. Communication in Mass Spectrometry

Rainfall interception in three contrasting lowland rain forest types in Central Kalimantan

A study of root water uptake of crops indicated by hydrogen and oxygen stable isotopes: A case in Shanxi Province

Correction of flux measurements for density effects due to heat and water vapour transfer

Diverse functional responses to drought in a Mediterranean-type shrubland in South Africa

Seasonal variations in moisture use in a pinon-juniper woodland

Water extraction times for plant and soil materials used in stable isotope analysis

A comparison of methods for determining forest evapotranspiration and its components: Sap-flow, soil water budget, eddy covariance and catchment water balance

The hydrology of the humid tropics

Soil and belowground characteristics of Pasoh Forest Reserve

Pasoh: Ecology of a lowland rain forest in southeast Asia

Seasonal variations in depth of water uptake for a subtropical coniferous plantation subjected to drought in an East Asian monsoon region

Organic carbon, nitrogen and mineral nutrient stock in the soils of Pasoh Forest

Evapotranspiration and water source of a tropical rainforest in peninsular Malaysia. Hydrological Processes