Saul A.Antezana-Vera·Ricardo A.Marenco

Abstract Minquartia guianensis Aubl.is a slow-growing species with several uses.In the juvenile state,it is well-adapted to low light conditions of the forest understory.However,it is still unknown how climate variability affects transpiration of this species,particularly under drought stress.In this study,we aimed to assess the effect of climatic variability on sap flow rates (SFR).SFR and radial growth were measured in six trees (14-50 cm diameter) in 2015 and 2016.Climate (precipitation,irradiance,relative humidity and temperature) and soil water content(SWC) data were also collected.SFR tended to increase in the dry season,with a negative relationship between SFR and SWC and precipitation (p <0.001),while there was a positive association between radial growth and monthly precipitation (p =0.004).Irradiance and temperature were the environmental factors more closely correlated with SFR during daytime (p <0.001),whereas relative humidity and vapor pressure def icit were the most important factors at night (p <0.001).Although negative SFR were sometimes recorded at night,the mean nocturnal sap flow was positive and across trees the nighttime sap flow accounted for 12.5%of the total daily sap flow.Increased transpiration during the dry season suggests that the root system of Minquartia was able to extract water from deep soil layers.These results widen our understanding of the ecophysiology of Amazonian trees under drought and provide further insight into the potential effect of the forecasted decline in precipitation in the Amazon region.

Keywords Reverse sap flow·Soil water content ·Transpiration·Tree growth

Abbreviations

ASActive sapwood area

DFDaily flow per tree

log10Base 10 logarithm

MGR Monthly diameter growth rate

PAR Photosynthetically active radiation

RH Relative humidity

RHmaxMaximum RH

RHminMinimum RH

SFR Sap flow rates

SWC Soil water content

TTemperature (°C)

TmaxMaximumT

TminMinimumT

TmeanMeanT

VPD Vapor pressure def icit (atmospheric)

VTTotal sap flow per tree (volume of sap)

Introduction

The Amazon rainforest is the largest tropical forest in the world and is particularly important because of its role in water and carbon cycling and its large biodiversity.It is estimated that the Amazon forest stores 86 Pg of carbon in total biomass (Saatchi et al.2007) and that approximately 50% of precipitation is recycled through transpiration (Salati 1987).Leaf cuticles are impermeable to carbon dioxide but permeable to water vapor,whereas photosynthesis depends on stomatal opening,and thereby is intrinsically associated with the loss of water through transpiration.Stomatal conductance typically declines with increasing atmospheric vapor pressure def icit (VPD) or with decreasing soil water potential (Jones 1998;Oliveira and Marenco 2019).However,during the drought of 2005,Saleska et al.(2007) reported an increase in ‘greenness’ and concluded that the gross primary productivity of the Amazon increased in the dry season.Restrepo-Coupe et al.(2013) also concluded that gross primary productivity increased during the dry season across the equatorial Amazon.On the other hand,Lee et al.(2013) and dos Santos et al.(2018) reported a decline in carbon uptake during the dry season in the central Amazon.A decline in photosynthetic rates may be associated with a decline in stomatal conductance (dos Santos et al.2018),and eventually in transpiration rates,particularly taking into account that there is often a decline in soil water content (SWC) during the dry season,in several parts of the Amazon (Malhi et al.1998;Juárez et al.2007;Barros et al.2019) .In fact,it has been found that,under water stress,transpiration of Amazonian trees can decline up to 30% under induced drought(50% reduction of incident precipitation,Costa et al.2018).However,it has also been reported that under water stress a slight increase (8—14%) in evapotranspiration can occur in the central Amazon (Hasler and Avissar 2007;Juárez et al.2007).Thus the real effects of drought on transpiration rates in central Amazonia are still to be clarif ied.

The possibility of foliar water uptake has interested researchers for centuries.For this phenomenon to occur,it is required an air-to-leaf water potential gradient,as well as the presence of anatomical structures capable of absorbing significant amounts of water (Rundel 1982),which can lead to reverse (shoot to root) sap flow in foggy environments(Goldsmith et al.2013).Although fog events can occur in the Amazon (Camargo and Marenco 2017),the contribution of fog to the foliar water balance of species in the Amazon seems to be negligible (Bastable et al.1993).

Minquartia guianensisAubl.(Olacaceae;hereafterMinquartia) is widely distributed in the Neotropics and can be found from Central America to the Amazon region(Hunter 1991).This slow-growing,shade tolerant species is well-adapted to low-fertility acid soils and high rainfalls,and in the juvenile state,the species grows in the dimly lit conditions of the forest understory (Marenco and Vieira 2005;Marenco et al.2017).However,Minquartiadoes not thrive under prolonged drought or on swampy soils (Hunter 1991).The tree often reaches 23-30 m in height and up to 90 cm in diameter,and its annual radial growth is approximately 1.3 mm (Hunter 1991;Dias and Marenco 2016).Minquartiahas a wide range of uses both in the wood industry and in traditional medicine.Local Amazonian people have usedMinquartiaas a fish poison and to treat several illnesses,including malaria,leishmaniasis,a parasitic disease,helminth parasites,tuberculosis,colds,muscular pain,skin irritations,and to relieve itching (Marles et al.1989).The antiprotozoal activity ofMinquartiaextract against leishmaniasis (Gachet et al.2010) and malaria has been demonstrated experimentally (Ruiz et al.2011).

Minquartiahas an irregular and rather fluted bole which limits its application in the timber industry.Nevertheless,because of its remarkable resistance to decay,termite attack,and wood-destroying organisms,the wood has been used for flooring and bridge construction,and in the production of railroad sleepers,fence posts,and poles (Hunter 1991).In fact,because of its wood properties (hard,heavy,strong,and resistant to decay),Minquartiahas been subjected to high logging rates and is currently listed as“near-threatened”in the IUCN Red List (IUCN-ARW 1998).This study is important because more frequent droughts have been predicted for the Amazon (Fu et al.2013) which may have negative impacts onMinquartia,a species typical of wet environments.We selected this species for study because of its importance in the Amazon.Although the photosynthetic traits have been reported elsewhere (e.g.Marenco and Vieira 2005;Marenco et al.2017),there is a scarcity of data on water relations of this species under field conditions in the Amazon region.Thus,in this study we aimed to assess the relationship between climatic variability and sap flow rates (SFR),and particularly the effect of drought on SFR ofMinquartia.We hypothesized that SFR decreases in the dry season in response to a decline in soil water content;we also hypothesized that SFR increases with increasing irradiance and VPD.

Materials and methods

Study area and plant material

The study was conducted on a plateau area of the ZF2 Experimental Station of the National Institute for Research in the Amazon (INPA),located 60 km north of Manaus(02° 36′ 21″ S;60° 08′ 11″ W).The region has a humid climate with a mild dry season from June to September (50 to 100 mm of precipitation per month in the dry season).The average annual temperature is 26 °C and the mean annual precipitation 2420 mm (Malhi and Wright 2004).In the dry season,VPD increases and canopy water content tends to decrease (Lee et al.2013).The soil is a yellow latosol of clay texture,pH of 4.2 to 4.5 and poor in nutrients.Data were collected in 2015 and 2016 from sixMinquartiatrees with a mean diameter of 23.4 cm,height of 23.6 m,crown diameter of 3.52 m,and wood density of 0.84 g cm−3(Table 1).

Table 1 Tree height,bole diameter at breast height (1.3 m),crown diameter,monthly diameter growth rate (MGR),wood density,sapwood depth,sapwood area,and total cross-sectional area (TCA) of the studied trees

Sap flow rate measurements

Sap flow rates were assessed using the heat ratio method(Burgess et al.2001).Data were collected for 268 days (average of 44.6 days per tree).To measure SFR two probes were placed upstream and downstream from a line heater (SFM1,ICT International,Armidale,NSW,Australia).Each probe consisted of a 1.3 mm (diameter) × 35 mm long needle with two thermistors at 7.5 and 22.5 mm from the needle tip.The probes remained in the same position for up to 30 days to avoid reducing their efficiency due to excessive wood injury.After this period,the probes were removed and reinstalled if necessary.The SFM1 instrument requires that the probes be perfectly aligned.Thus,if necessary (e.g.,slight misalignment),the probe was carefully removed and reinstalled.SFR data were collected at 30-min intervals;to power the sap flow meter a 12 V (60 A) battery was used.

Sap velocity describes the sap flow per unit sapwood area over time,hence sap flow in rate units (SFR in cm3(g of water) cm−2h−1) is used instead of sap flow in centimeters per hour.SFR was calculated according to Burgess et al.(2001),with modifications proposed by Vandegehuchte and Steppe (2012).

The total volume of sap flow per tree (VT) was also estimated as the product of sap flow rates (SFR) and the crosssectional area of active sapwood area,AS(Burgess et al.2001).Daily flow per tree (DF) was obtained by integrating theVTvalues at 30 min intervals (VTi) over time.The sapwood depth was estimated as the intercept of the regression line of the maximum SFR (inner and outer probe) against the thermistor depth,i.e.,at SFR of zero (an illustration is provided in the result section).We assumed that,beyond the outer thermistor,the sap velocity decreased linearly to zero at the heartwood.

Tree growth and wood properties

To determine the sapwood properties,a sapwood core was extracted from each tree with a 5.15-mm internal diameter increment borer (Haglöf Sweden,Langsele,Sweden).After extraction,the sample was placed into a small capped test tube,and the sapwood fresh volume and the dry and fresh mass determined.The dry matter was obtained after oven drying at 80 °C to constant mass.These core samples were used to determine wood density,sapwood water content and thermal diff usivity (Vandegehuchte and Steppe 2012).

Radial growth at breast height (1.3 m from the ground)was measured at monthly intervals over 24 months(2015 − 2016) using stainless steel dendrometer bands which had been installed three years before the beginning of the study.In addition,in some trees with a sap flow measuring probe,the variation of radial growth over time was also measured with logging band dendrometers (DRL26,Environmental Measuring Systems,Brno,Czech Republic)installed at breast height.

Climatic data

Climatic data was collected at a 40-m tall observation tower(02°35′21″ S,60°06′53″ W) during the study period.Air temperature (T),relative humidity (RH),and photosynthetically active radiation (PAR) data were collected at 15 − 30 min intervals with specific sensors (Humitter 50y,Vaisala,Ov,Finland;LI-190SA,Li-Cor,Lincoln,NE,USA)connected to a data logger (Li-1400,Li-Cor,Lincoln).Daily precipitation data were collected with a tipping bucket gaugeconnected to a data logger (ECR-100,Em5b,Decagon Devices,Pullman,WA,USA).A second conventional rain gauge was installed in the same tower to collect data at oneweek intervals in order to generate a second data set in case of technical difficulties with the tipping bucket gauge.Vapor pressure def icit (VPD) was calculated from RH and temperature data as described by Marenco et al.(2014b).We also collected undisturbed soil samples (100 to 200 mm depth)at two-week intervals to determine soil water content after drying the samples at 105 °C.To correlate soil moisture with daily sap flow rates,the SWC data set was complemented with data compiled from NASA for the coordinates of the study site (GLDAS_NOAH025,https://giova nni.gsfc.nasa.gov/giova nni/).

Data analysis

Regression analysis was used to describe the relationship between sap flow rates and SWC and between SFR and precipitation,and also to examine the association between tree growth and precipitation.Pearson correlation was used to determine the relationship between SFR and VPD,PAR,temperature and RH.Growth and SFR data were log10-transformed prior to statistical analysis.SigmaPlot 11.0 (Systat Software Inc.San Jose,CA,USA) was used for data analysis.

Results

During the study period,the mean temperature (Tmean) was 26.4 °C and the mean minimum (Tmin) and mean maximum(Tmax) temperatures were 23.3 °C and 31.1 °C,respectively (Fig.1 a).Annual precipitation was 1983 mm in 2015 and 2021 mm in 2016 with a monthly precipitation that ranged from 23.2 mm in September to 477 mm in March(Fig.1 b).Mean maximum and mean minimum RH were 90.7% and 53.8%,respectively (Fig.1 c),with an average of 76.0% throughout the experimental period.The mean values of VPDmin,VPDmeanand VPDmaxwere 2.7,8.5 and 21.4 hPa,respectively,and the daily PAR ranged from 17.0 to 39.7 mol m−2d−1,with a mean of 26.4 mol m−2d−1.Average soil water content was 42.3% (v/v),with a range of 32.7% to 52.0% (Fig.1 c).

Fig.1 Climate parameters and soil water content (SWC,% v/v)recorded between January 2015 (month 1) and December 2016(month 24) at the study site:a temperature (T:mean,minimum -min and maximum -max);b precipitation;c photosynthetically active radiation (PAR,mol m−2 d−1),vapor pressure def icit (VPD,hPa),relative humidity (RH,%) and SWC;d monthly diameter growth rate(MGR,mean ± SE) of Minquartia guianensis.For temperature,RH,VPD and PAR each symbol denotes the mean of one month,for precipitation the symbol denotes the precipitation accumulated over the indicated period of time

Sap flow rates

During the daytime,PAR and temperature were the environmental factors closely correlated with SFR (p<0.001,Table 2,Fig.2).Therefore,the greater sap flow rates were observed on clear,sunny days and the lower ones on overcast days,i.e.,when PAR and temperatures were lower (Fig.2).Besides PAR,VPD also showed a positive correlation with SFR during the daytime (r≈ 0.50,Table 2,p<0.001).On the other hand,SFR decreased with increasing RH (Table 2).The outer probe was slightly more responsive (higherrvalues) to climatic parameters than the inner probe (Table 2).

Fig.2 Diurnal variations in SFR,VPD (a) and PAR (b) recorded in Tree number 16 between January 24 and February 04 of 2015.Acronyms:SFR,sap flow rates;PAR,photosynthetically active radiation,and VPD,vapor pressure def icit

At night,RH and VPD were the environmental factors most closely correlated with SFR,and within a given probe,these parameters showed similar absolutervalues (Table 2).The correlation between nocturnal SFR and temperature was lower than that observed in daytime.This occurs because the correlation between temperature and the inner probe SFR was very low and insignificant at night (p=0.21,Table 2).The average daily SFR increased with a decline in SWC,and also with a decrease in precipitation (Fig.3),but although significant,the regression model between SFR and precipitation (or SWC) only explained 4-10% of the total variance(Fig.3).

Fig.3 Relationship between the daily mean sap flow rate (SFR) and soil water content,SWC (a) and between SFR and precipitation (b).SFR data were log10-transformed prior to statistical analysis.Each symbol represents the daily mean SFR (mean of 48 points collected daily at 30-min interval)

The maximum average SFR recorded by the inner and outer probes were 9.2 and 25.0 g cm−2h−1,respectively(Table 3),and over a 24-h period,the average daily SFR values were 3.2 g cm−2h−1(inner probe) and 6.5 g cm−2h−1(outer probe),with an average SFR of 5.8 g cm−2h−1across the whole sapwood ring (Table 3).The mean daily SFR results in a root-canopy sap transport time of 16.8 days(mean tree height of 23.6 m and mean sap flux velocity of 1.4 m day−1).Except for tree # 390,most of the time the outer probe recorded higher sap flow rates than the innerprobe (Fig.4).In tree # 390,both probes tended to record similar maximum SFR values (Table 3).Mean maximum nocturnal sap flow rates recorded by inner and outer probes were 2.2 and 3.7 g cm−2h−1,respectively (Table 3).In most instances,nighttime SFR was positive but negative sap flow rates were not entirely absent during the study period(Table 3).Irrespective of rainfall conditions (wet or dry season),the reverse flow phenomenon was only observed at night.It was more often recorded by the inner probe and often at rather low intensity (absolute value lower than 2.0 g cm−2h−1,Table 3).On average,the nighttime SFR was 12.5% of the total sap flow per tree,while the mean nighttime/daytime ratio was 14.3% (Table 3).

Table 2 Correlation coefficients between the inner and outer maximum SFR and the climatic parameters

Sapwood area and tree growth

The sapwood ofMinquartiawas shallow (2.9-cm depth,as illustrated in Fig.4 b),which resulted in a mean sap wood area of approximately 200 cm2per tree (Table 1).Furthermore,the sap wood area to total cross-sectional area ratio varied considerably among trees,from 25% (in the largest tree) to about 50% in the smaller ones,with a mean of 34.9%(Table 1).The logging band dendrometer showed that often there was a slight bole shrinking (5—7 μm) toward midday,particularly on sunny days,followed by a slight swelling toward midnight (Fig.5 a).Radial growth increased with increasing precipitation (r2=0.31,Figs.1 d,5 b) and throughout the study period,there was an average monthly diameter growth rate of 0.106 mm per month (Table 1).

Fig.4 Diurnal sap flow rates (SFR) recorded by the outer (a) and inner (b) probe in Tree number 157 between January 16 and January 28 of 2015 (please note the different y-scale in the panels).The inset in panel B illustrates the procedure used to estimate the sapwood depth (SWD,in cm) at zero SFR (blue star),and the diamonds show the outer and inner maximum SFR at 0.95 and 2.45 cm SWD

Fig.5 Diurnal variation in tree diameter and diurnal course of photosynthetically active radiation (PAR,a),and the relationship between monthly diameter growth rate (MGR) and precipitation (b) during the study period.Data of panel A were recorded in Tree number 16 between January 24 and February 04 of 2015.MGR data were log10-transformed prior to statistical analysis

Discussion

In 2015 and 2016,mean annual precipitation was 2002 mm,below the historical mean of 2420 mm for central Amazonia(Malhi and Wright 2004).The soil water content (SWC) in the dry season (32—40%) was lower than the soil moisture content of 42-45% reported by Malhi et al.(1998).In a yellow latosol of this region,the permanent wilting point,PWP(tension of − 1.5 MPa),often occurs at a SWC of 0.30 v/v(Ranzani 1980),which indicates that during the dry season (particularly in 2015),the upper soil layer was close to the PWP.The SWC values in this study are in agreement with those found by Barros et al.(2019) at a site 8.5 km away (60.21° W,02.61° S).They reported a decline in SWC (51—42%),an increase in the cumulative water def icit(200-400 mm),and an increase in VPD with progressive drought during the prolonged dry season of 2015.Precipitation intensity was below 100-110 mm month−1for several months.In fact,the dry season of 2015 was more prolonged than that of a typical year as it extended to January 2016(Fig.1).

The maximum SFR recorded in this study is within the range of values (5-70 g cm−2h−1) reported in a study of 125 Neotropical trees (Kunert 2016).Whereas the mean daily SFR observed in this study (5.82 g cm−2h−1) is higher than the average value (2.9 g cm−2h−1) recorded by O’Brien et al.(2004) forMinquartia guianensisin a rainforest site(mean VPD of 0.7 kPa) with no marked dry season and 4000 mm year−1precipitation.Also,our mean daily SFR is higher than the SFR found with other tropical species(1.55-4.70 g cm−2s−1) such asCryptocarya laevigataBlume,Pouteria firma(Miq.) Baehni,Platea excelsaBl.var.borneensis(Heine) Sleum.andCastanopsis acuminatissima(Blume) Rheder in a perhumid site (average VPD of 0.4-0.5 kPa) with a short dry season and 3500 mm year−1precipitation (Horna et al.2011).Thus,the discrepancy with these reports can be ascribed to the prevailing climatic conditions during data collection,as this study report data collected under drier conditions with high temperatures and low humidity associated with high VPD values.

The greater SFR recorded by the outer probe shows that the outer sapwood ring has greater hydraulic conductivity than the innermost sapwood (Zhao et al.2018).This may be related to the process of heartwood formation (Taylor et al.2002).As an exception,in Tree # 390 both the inner and the outer probes recorded similar SFR values because,for no clear reason,the outer probe recorded low sap flow values.The negative correlation between SFR and SWC shows that the trees tend to lose more water in the dry season,as suggested by Hasler and Avissar (2007).Juárez et al.(2007) also found that,in the central Amazon,the evapotranspiration rate was slightly higher in the dry season (3.43 mm day−1over 3.17 mm day−1in the wet season),despite the drop in stomatal conductance in the dry season (dos Santos et al.2018).

Irradiance,temperature and VPD were the climatic parameters more closely correlated with SFR during daytime.This is in agreement with O’Brien et al.(2 004) and Motzer et al.(2005) who observed that the highest SFR occurred around midday.Zhao et al.(2017) also found that inPopulus euphraticaOliv.SFR correlates positively with irradiance (r=0.77),temperature (r≈ 0.90) and vapor pressure def icit (r≈ 0.90).Our data also agree with the results reported by Horna et al.(2011) who found a lower correlation between SFR and SWC than between SFR and irradiance.This can occur because sap flow is greatly affected by weather conditions such as cloudiness.We found that SFR positively correlated with temperature and VPD and hence negatively associated with RH.The effect of VPD on SFR occurs because the difference in vapor pressure between a leaf and its surrounding leads to a negative pressure (tension) within the xylem conduits,the ultimate driver of transpiration (Tyree and Zimmermann 2002).Thus,temperature and RH can indirectly affect SFR via their effect on VPD.It is difficult to separate the effects of the individual components of VPD on leaf conductance and transpiration.An increase in temperature directly affects the metabolism of guard cells,which can lead to an increase in stomatal conductance and transpiration (Jones 1998;Kostaki et al.2020).Besides,the conductivity of the xylem conduits can increase with increasing temperature.Temperature has a direct effect on transpiration via the effect of temperature on water viscosity (Darcy’s Law),membrane permeability of the cell,and cuticular transpiration -peristomatal transpiration (Burghardt and Riederer 2006).In this study,we measured mean maximum temperatures of 30—35 °C and Brinkmann et al.(1971) measured (July 1968-July 1969)soil temperatures of 23—25 °C at a Forest Reserve under forest cover near Manaus (central Amazonia).A 10 °C-gradient between soil and shoot temperatures may increase the time lag between water uptake by roots and water loss in the upper canopy.SFR reached a peak near midday and followed the diurnal variation in PAR and VPD.Transpiration greatly depends on stomatal conductance,and via a complex feedback mechanism,stomata also respond to changes in transpiration rates (Jones 1998).In a previous study,it was found that,under constant irradiance,photosynthesis and stomatal conductance peaked at about 10:30 in a 26-m-tall tree (Marenco et al.2014a),that is,about 60-90 min earlier than the SFR peak time recorded in this study.A time lag between the peak in canopy transpiration and the peak in stem SFR can be attributed to the release of stored water from branch and stem tissues -tree capacitance (Schulze et al.1985,Goldstein et al.1998).

The primary roles of stomata include minimizing water loss per unit of assimilated carbon (Farquhar and Sharkey 1982),and controlling water loss to avoid the risk of xylem cavitation (Jones and Sutherland 1991).Thus,it is unlikely that the decline in SFR after midday is a direct response to a reduction in PAR.Instead,it is more plausible to infer that the decline in SFR after midday is a mechanism to reduce the risk of embolism (Jones and Sutherland 1991;Jones 1998).Although xylem repair under tension can occur in small trees (Brodersen and McElrone 2013),xylem repair in tall trees still remains to be demonstrated (Koch and Fredeen 2005).

There is considerable evidence to support the suggestion that plant transpiration often decreases when the difference between soil water potential and leaf water potential becomes progressively smaller (Jones 1998).Thus,one can expect that a decline in SWC during the dry season should lead to a decrease in SFR under water stress.We hypothesized that SFR should decrease in the dry season,which was not supported by data because the SFR tended to increase with decreasing precipitation and SWC.In this respect,our results concur with those reported by Hasler and Avissar(2007) who found that in the dry season,the latent heat flux(evapotranspiration in energy units) increased by 13-14%in the dry season in the central Amazon.An increase in transpiration rates during the dry season may ref lect the ability of the trees to extract water from the deeper layers of soil,as suggested by Juárez et al.(2007).Most commonly,the nocturnal SFR was positive,which shows that nighttime transpiration represents an important component of the tree water balance.This is in agreement with the results reported by Zhao et al.(2019).They found that,under arid climatic conditions,the nocturnal transpiration ofPopulus euphraticaOliv.can represent 15% of daytime transpiration.Negative sap flow rates were sometimes observed at night,which was not expected to occur under drought.The reverse sap flow was more prevalent in the inner sapwood,which suggests that,inMinquartia,the inner and outer sapwood ring are not entirely integrated.

It is well-known that VPD and soil water potential are the driving forces of transpiration.Thus,transpiration will cease when either VPD is zero or when the difference between leaf water potential and soil water potential is null (van den Honert’s law).Sap flow from roots to dry soil layers has been observed in several environments and species (Domec et al.2010) .For reverse flow to occur,it is crucial to relieve the pressure gradient within the xylem conduits,which may involve foliar water uptake (Goldsmith et al.2013).This implies that the air surrounding the leaves must be very close to the saturation point,as the water potential of the atmosphere is very sensitive to changes in air humidity.RH values very close to 100% were often recorded (Fig.1 c).Thus,it seems possible that a water—vapor saturated atmosphere in the tree canopy contributed to the occurrence of reverse sap flow.

The radial growth reported in this study(0.106 mm month−1) is barely within the lower range of values (0.10 to 0.4 mm month−1) reported in a previous study involving other 27 tree species (Dias and Marenco 2016).Low radial growth rates during the dry season can be attributed to a decline in stomatal conductance (dos Santos et al.2018),which ultimately leads to low photosynthetic rates(Lee et al.2013).We noted that the diameter of a tree can shrink during daytime,which can be attributed to an increase in VPD during the hottest hours of the day.Radial swelling and shrinking is not an uncommon phenomenon in tropical trees,which can contribute to negative variation in tree diameter during the dry season (Dias and Marenco 2016).

Conclusions

We hypothesized that during the dry season a decline in soil water content (SWC) reduces sap flow rates (SFR).This was not supported by data because SFR did not decline in the dry season,instead it tended to increase.Nevertheless,the determination coefficient of the linear regression between SWC and SFR was very low,which suggests caution in establishing a cause-effect relationship between drought and wholetree transpiration.Our hypothesis on the effect of irradiance and VPD on SFR was supported by data,since both photosynthetically active radiation (PAR) and VPD were correlated with SFR.Although the SFR tended to increase in the dry season,tree growth declined during the drier months which can be the result of a decline in stomatal conductance.The mean nighttime sap flow was positive,which indicates that the nocturnal transpiration represents an important component of the tree water balance.The occurrence of negative SFR at night during the wet and dry season suggests that this phenomenon is not directly related to SWC but to the conditions prevailing in the canopy—atmosphere environment.These results widen our understanding of the ecophysiology of Amazonian trees under water def icit and shed light into the potential effects of the ongoing climate change on the Amazon region,particularly those related to spatial and time distribution of precipitation.

AcknowledgementsThanks to the Ministério da Ciência,Tecnologia e Inovações (MCTI-INPA,PRJ-15.120),Conselho Nacional de Desenvolvimento Científ ico e Tecnológico (CNPq,fellowship to RAM—303907/2018-5,and scholarship to SAAV),Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES,code 001)and Fundação de Amparo a Pesquisa do Amazonas (FAPEAM) for supporting this research.