Estimating plant crown transpiration and water use efficiency by vegetative reflectance indices associated with chlorophyll fluorescence

This research developed estimates of plant crown transpiration and water-use-efficiency using reflectance and derivative indices extracted from remotely sensed chlorophyll fluorescence measurements under natural conditions. Diurnal changes of leaf-level gas exchange (carbon assimilation rate (A), stomatal conductance (gs), transpiration rate (E), chlorophyll fluorescence and canopy-scale remote sensing were measured on top crown of valley oak (Quercus lobata) in the foothills of central California, USA. The results indicated Q. lobata experienced saturating irradiance (PAR), which induced photoinhibition indicated by a decrease in the quantum efficiency of photosystem II (r = 0.648 with   v m F F and r 2 = 0.73 with PSII) and open reaction centers (qP; r = 0.699). The excess absorbed quantum energy was dissipated as heat through the Xanthophyll cycle and other processes (photorespiration and the water-water cycle) rather than energy emission as steady state chlorophyll fluorescence (Fs). An increase in leaf temperature caused by the activity of Xanthophyll cycle was correlated to a decrease in Fs (r = 0.381) and an increase in evaporative cooling through E (r = 0.800) and water use efficiency (WUE; r = 0.872).


INTRODUCTION
Estimation of plant crown transpiration and wateruse-efficiency (WUE) based on remotely sensed vegetative indices has the potential to enhance the study of ecosystem water flux and how species and ecosystems could respond to future climatic induced water stresses.In their Fourth Assessment Report, the Intergovernmental Panel on Climate Change reported that climate change has induced variation in precipitation patterns globally during the last century; some regions such as eastern parts of Northern and South America, northern Europe and northern and central Asia have increased their precipitation while other areas such as the Shale, the Mediterranean, southern Africa and southern Asia have had reductions [1].Long-term droughts have been observed and have affected agriculture and economic development in some semi-arid and sub-humid regions of the globe including the western U.S. [2].It is projected that some of current water stressed areas will experience even more severe drought and an increase in the frequency of drought during the 21 st century [2,3].
Because the impacts of water stress vary in time, across space and between species, they generate shifts in species abundance of forest vegetation [4].Water stress also changes vegetative water use patterns, such as WUE.These water stress impacts on vegetation gas exchange play a very significant role in the local to global carbon cycles [5,6].Therefore, environmental stresses such as, heat, fire and insect stresses or vegetation water status have become the subject of remote sensing studies because spectral indices of vegetation can be informative about plant and ecosystem physiological conditions in-cluding atmospheric-terrestrial gas exchange processes.
Spectral reflectance indices commonly used for remote sensing are dependent on photosynthetic pigment concentration and plant water content [6][7][8][9][10][11].In addition to these indices, reflectance indices associated with chlorophyll fluorescence have been successively used to estimate photosynthetic activities under heat and drought stress, chlorophyll fluorescence itself is linked to physiological stress in plants [12][13][14][15][16][17].Currently, remote estimation of chlorophyll fluorescence has been proposed to integrate the physiological function at the ecosystem scale projecting net primary productivity (NPP).Quantum yield of PSII  v m  F F   and the steady-state fluorescence (F s ) have been proposed for estimating the photosynthetic radiation use efficiency at large scales, suggesting remote sensing of chlorophyll fluorescence parameters as a tool for large scale CO 2 flux and transpiretion measurements [12].
In addition to the use of chlorophyll fluorescence for CO 2 flux measurements, Flexas et al. (2002) [18] demonstrated that F s tracked stomatal conductance (g s ) rate of field-grown grapevines under drought conditions.However, this experiment was conducted at a single-leaf scale, not at a larger scale such as canopy, stand or ecosystem scale.The usefulness of chlorophyll fluorescence and remote sensing of chlorophyll fluorescence for largescale transpiration is still unclear.This study is an investtigation of physiological responses to environmental stimuli based on seasonal and diurnal observational measurements including micrometeorological data, gas exchange, chlorophyll fluorescence, and remotely sensed reflectance and derivative indices.We suggest means to estimate crown scale transpiration and WUE of a field grown oak tree Quercus lobata (valley oak) using spectral reflectance indices associated with chlorophyll fluorescence.

Study Site
This experiment was conducted in Big Chico Creek Ecological Reserve (BCCER).BCCER is owned and managed by the Research Foundation of California State University, Chico for the purpose of preserving the critical natural habitat and providing environmental research and educational areas [19].It is located in the foothills of the Sierra Nevada in the northern portion of the Sacramento Valley about 10 miles northeast of Chico, California, USA.The Reserve ranges in elevation from 213 to 623 feet, with mean precipitation ranging from 64 cm in the valley to 203 cm in the headwater region with hot dry summers and extended periods of limited rainfall.The Reserve includes 7.24 km of Big Chico Creek and encompasses 1599 ha of land.(Latitute 39˚51'51''N Longi-tude 121˚42'46''W).

Plant Material
A naturally occurring Quercus lobata (valley oak) with an open crown was selected for the study plant.This oak tree was found in a riparian area, which was located in the canyon of Big Chico Creek, and about 30 m from the creek (Figure 1).The tree had about 90 cm DBH and approximately 20 m height.The measurements including gas exchange, chlorophyll fluorescence and spectral reflectance were conducted on the southwest portion of the crown utilizing a constructed tower at approximately 15 m height.

Meteorological Measurements
A micrometeorological station was established about 40 m south of the study tree and 30 m from Big Chico Creek for the purpose of monitoring micro meteorological data including solar irradiance, air temperature and relative humidity at 2 m height from the ground (Figure 1).A quantum sensor [20] was mounted on the micro meteorological station.This sensor read global radiation (W•m −2 ).The station was equipped with a humidity probe (HMP35A, Waisala Inc., Helsinki, Finland).The probe read relative humidity in percent and a thermistor (UUT51J1, Fenwal Electronics, Toledo, Ohio) was used to observe air temperature in degree Celsius.To utilize the recorded data for this research, specific data (from 10:30 to 16:30 of experiment days) were read out to Microsoft Excel spread sheets.Time data of gas exchange, chlorophyll fluorescence and spectral reflectance measurements discussed below were modified to match with time data of weather data.

Gas Exchange and Chlorophyll Fluorescence Measurements
Leaf level gas exchange and chlorophyll fluorescence were measured using LI-COR 6400 infra-red gas analyzer (IRGA) [14] equipped with a leaf chamber fluorometer [20].Diurnal cycles of gas exchange and chlorophyll fluorescence were taken from 10:30 to 16:30 on September 29 th , October 6 th , 7 th , 28 th and November 4 th .Intact leaves of the top crown were selected at the beginning of the measurements each day.These leaves were equilibrated prior to measurements.The LI-COR 6400 was programmed to take a sequence of measurements at 20 minute intervals of gas exchange and chlorophyll fluorescence.Gas exchange measurements included carbon assimilation rate (A, µmol CO 2 m −2 •s −1 ), stomatal conductance (g s ), (µmol  [22].

Reflectance Measurements
Crown reflectance of Q. lobata was taken with a portable spectrometer (UniSpec-SC, PP Systems, Haverhill, MA).A 2.1 mm diameter fiber optic with an approximate field of view (FOV) of 20˚ was mounted to a measurement station at 1 m height from the top of the crown at the nadir angle of 90˚.The measurement FOV at crownheight was 0.977 m 2 .
The reflectance measurements were taken simultaneously with gas exchange and chlorophyll fluorescence measurements (every 20 minutes between 10:30 to 16:30).Between the measurements, Unispec-SC was calibrated using a reflectance-standard panel (Spectralon, Labsphere, New Hampshire) and the dark current was corrected for each measurement.Five readings (each composed of 40) scans of reflectance were taken and then averaged.The data were processed with software, Multispec, that interpolated from the original waveband (3.3 nm) of Uni-Spec-SC to 1 nm intervals to yield crown reflectance and derivative spectra and to calculate reflectance and derivative indices.

Data Analysis
Quadratic regression using a second order polynomial was used to determine relationship between two different measurement variables in this study.Calculated relation values (r 2 and p-value) were used to predict a Y-value from an X-value.The data were analyzed with polynomial regression function in scientific data analysis software (SigmaPlot ver.10.0, Systat Software Inc., San Jose, CA).

Micrometeorological Data
Micrometeorological data revealed a typical pattern of seasonal diurnal cycle.Solar irradiance generally showed a gradual increase before noon and reached its peak between noon and 13:00 (Figure 2(a)).It started decreasing around 14:00 to 15:00.A sudden reduction was observed around 16:00 due to the landscape of the canyon blocking the sunlight shading the quantum sensor.Over the experimental period, a mean solar irradiance (mean ± standard error) dropped from 921.4 ± 46.9 W•m −2 (September 29 th ) to 567.1 ± 54.3 W•m −2 (November 4 th ).The lowest mean and highest global radiations were observed on October 28 th (562.2 ± 46.0 W•m −2 ) and September 29 th , respectively.
Air temperature and relative humidity followed the diurnal cycle of solar irradiance (Figures 2(b) and (c)).However, the daily mean air temperature slightly increased from 20.4˚C ± 0.5˚C on September 29 th to 23.9˚C ± 1.1˚C November 4 th .The daily mean relative humidity didn't change (28.8% ± 1.9% to 28.3% ± 3.5%).The highest mean air temperature (25.8˚C ± 0.6˚C) and relative humidity (34.6% ± 2.0%) were observed on October 28 th .
Figure 4 shows time course measurements of WUE and LUE.

1
v m  F F q     P or light energy that was not converted to photochemical energy thermally dissipated increased between 1000 and 1500 µmol photon m −2 •s −1 of PAR and then stopped increasing its allocation at PAR greater than 1500 µmol photon m −2 •s −1 (r 2 = 0.648).
  P stand for the fraction of absorbed light that is dissipated as heat, utilized for photochemistry (same as ФPSII) and excess light energy not dissipated in PSII antenna.The sum of all three fractions at each PAR level becomes 1 (100%).All data have P-value < 0.001 (n = 78).
decreasing at leaf temperature higher than 28˚C (r 2 = 0.381).E didn't show a significant change between 20 and 25˚C, but increased its rate at leaf temperature higher than 25˚C (r 2 = 0.800).
Light use efficiency calculated as A divided by PAR, and WUE, A/E, of Q. lobata inversely tracked changes in g s , PAR and leaf temperature (Figure 4).Both LUE and WUE increased their value as stomata started closing.At the same time, leaf temperature yielded its highest values of the day and PAR begun decreasing (Figure 2(d)).Mean and standard error of LUE and WUE were 0.276 ± 0.0055 μmol CO 2 m −2 •s −1 /μmol photon m −2 •s −1 and 0.0121 ± 0.00022 μmol CO 2 m −2 •s −1 /μmol H 2 O m −2 •s −1 , respectively.LUE was calculated based on the assumption of 84 % absorbance of the incident PAR by the leaf.

Remote Sensing Results
Daily mean spectra of crown reflectance and crown derivative are shown are shown in Figure 6.Reflectance and derivative spectra (Dpi) showed changes during the measurements reflecting a decrease in PAR due to the seasonal variance in the solar declination angle.Reflectance spectra exhibited an increase in a xanthophyll related peak centered on 550 nm, and a shift in the red edge toward shorter wavelengths.Derivative spectra revealed an increase in a peak at 700 nm (between 690 and 710), but no change was observed in the peak at 720 nm(between 710 and 730 nm).R740/R800 (r 2 = 0.292 for D705/D722, 0.265 for Regression analysis between fluorescence ratio indices (FRI), which are R690/R600 and R740/R800, and derivative chlorophyll indices (DCI) including D705/D722, D730/D706 and (D720-D703)/D707 was conducted to determine the linkage of DCIs to chlorophyll fluorescence emission spectra (Figure 7).All three DCIs were significantly correlated to R740/R800 (r 2 = 0.731 for D705/ D722, 0.692 for D730/D706 and 0.715 for (D720 − D703)/D707).DCIs were also significantly correlated to R690/R600, but showing poorer linkages than D730/ D706 and 0.198 for (D720 − D703)/D707).These results indicate DCIs are strongly associated with chlorophyll fluorescence emission at 740 nm (F740) rather than 690 nm peak (F690).The fluorescence ratio index (FRI R690/ R600) shows significant correlations with PAR, chlorophyll fluorescence parameters and the allocation of absorbed light energy by PSII antenna (Figure 8).R690/ R600 tracked PAR with a positive correlation (r 2 = 0.486) although it didn't correlate with leaf temperature change (Figure 2(d 8(d)).In addition, the index was found to be poorly correlated to physiological responses to leaf temperature changes (Figure 2(d)).Figure 9 shows a regression analysis of D705 /D722 indices.The index had a significant positive correlation to leaf temperature and physiological responses (r 2 = 0.292 for leaf temperature, 0.537 for E, 0.501 for g s and 0.712 for Ci/Ca) (Figures 9(a) to (d)).Figures 9(e) and (f) demonstrate a significant correlation between D705/ D722 and commonly used reflectance index: PRI (r 2 = 0.785) and water band index or WBI (r 2 = 0.595).The other two DCIs yielded similar correlations with these parameters (Table 1).FRI (R740/R800) was strongly correlated to E and g s while it was poorly correlated to leaf temperature, PRI and WBI (Table 1).The indices associated with F740 were found to have very weak   linkage to PAR and chlorophyll fluorescence Parameters (data not shown).

DISCUSSION
The diurnal cycle measurement of PAR (Figure 3) revealed that under our regional conditions Q. lobata received higher quantum fluxes than its photosynthetic capacity (500 µmol photon m −2 •s −1 ) (Figure 3).This indicates Q. lobata experienced excess light stress during the experiment and couldn't fully utilize all absorbed light energy for photosynthetic activity.Actually, only a small portion (1.51% ± 0.028%) of absorbed light energy was consumed by carbon assimilation in the Calvin cycle while 27.6% ± 0.55% of absorbed quanta was converted to photochemistry by PSII (Figure 6).Also, the result of the regression analyses of PAR with LUE (Figure 6), maximum quantum yield of PSII  v m  F F   and quantum yield of PSII (ΦPSII) (Figure 8(a)) indicate that less absorbed quantum energy was used in carbon assimilation and converted to photochemical energy as light intensity increased.The observed decrease in v m F F   has been suggested to be the result of degradation associated with the D1 protein [25].The increase in the excess light stress, v m F F   (1 − qP) (Figure 8(b)), has been shown to enhance the formation of damaging singlet oxygen in PSII [22] that is associated with the degradation of the D1 protein [26].At the same time, the fraction of the light energy dissipated as heat (1 − v m F F   ) had increased (Figure 8(b)).These results suggest that under excess light conditions, most of the quantum energy was dissipated through the xanthophylls cycle.
As a result of the observed increase in heat dissipation of light energy, heat stress was considered to have an inhibitory effect on the Calvin cycle [27] and the accumulation of NADPH that increased the reduced form of plastoquinone (PQ) molecules [28].Reduced PQ molecules could enhance the cyclic electron transport from ferredoxin to PQ and the proton gradient through the thylakoid membrane [29].These physiological mechanisms contributed to activate the energy-dependent quenching (qE) [30] and serve to protect PSI from superoxide ions under the high light condition [26].Joly and Carpentier (2007) [28] reported that the reduction of PQ molecules, caused by the accumulation of NADPH, was induced by the inhibition of Rubisco activase, and decreases the activation states of Rubisco under high leaf temperatures [27], enhanced the energy dissipation in PSI as chlorophyll fluorescence centered around 740 nm (F740-spectrum).
As leaf temperature increased during this experiment (Figure 2), F740-spectrum was demonstrated to increase, as the ratio of Ci/Ca increased suggesting a decrease in Rubisco activity.However, chlorophyll fluorescence emitted from PSII, which was centered around 690 nm (F690-spectrum), was believed to decrease under the high light conditions.These different responses between properties F690 and F740 spectra might be attributed to the measurement with LI-COR 6400-40 which was designed to detect fluorescence emission at 715 nm.Recently, the contribution of F740 spectrum to the total fluorescence spectra at wavelength greater than 700 nm was found to be significant [31][32][33].In this experiment the equipment design and chlorophyll fluorescence properties resulted in ecologically strong r 2 -correlations (0.64 to 0.73) between PAR and chlorophyll fluorescence parameters (Figures 7 and 8).Fluorescence ratio indices (FRI; R690/R600 and R740/R800), curvature index (R675 × R690)/R683 2 ) and derivative chlorophyll indices (DCI; D705/D722, D730/D706 and (D720 − D703)/D707) successfully yielded the remote estimation of the different patterns of F690 and F740 spectra in this experiment.Indices associated with F690-spectrum, R690/R600 and (R675 × R690)/R683 2 , were found to have linkage to chlorophyll fluorescence parameters measured with LI-COR 6400-40 (Figure 9 and Table 1).This result suggests the use of reflectance indices associated with 690 nm wavelength allows detecting changes in PSII fluorescence.This observation is supported by Zarco-Tejada et al. 2003a and 2003b [17,34].They reported the direct relationship between PSII-related chlorophyll fluorescence (F s , m F  and v m F F   ) measured with PAM-2000 Fluorometer and remote estimation of F690 in laboratory, the field and from aerial sensors under both light-saturating and solar-illumination.However, r 2 -correlation between these indices and chlorophyll fluorescence parameters yielded weak values.The reason for the weak r 2 -correlation is considered as following.The gap between the actual irradiance on the crown and the PAR inside the leaf chamber is considered as the cause of weak values of r 2 -correlation in addition to the contribution of F740-spectra to the total chlorophyll fluorescence and LI-COR 6400-40 design.In addition, an influence of a change in the quality of solar irradiance spectrum due to the seasonal change in the solar zenith angle can be considered to have caused a change in F690 It should be noted that the position of the quantum sensor was re-positioned on October 6 th and November 4 th due to setting problems and the sample leaf was changed because of accidental leaf break under gusty wind conditions.
It is considered that these problems reduced the accuracy of F s measurements because chlorophyll fluorescence required longer periods to stabilize than did gas exchange.This fact might have skewed the correlation between leaf-level F s and both gas exchange measurements and canopy-scale chlorophyll fluorescence.R740/ R800 and DCI, which were associated with F740-spectrum, were significantly correlated with leaf temperature, Ci/Ca, E, PRI and WBI (Figure 9 and Table 1).These reflectance indices successively cycle activity.The relationship between PSI fluorescence and crown transpiretion is not clear.However, both physiological activities are dependent on leaf temperature.These properties of PSI fluorescence and transpiration suggest reflectance indices associated with F740 can be used to estimate crown transpiration (Figure 9 and Table 1).In addition, remote estimation of F740 was found to be not directly dependent on variation in solar irradiance.This indicates the use of F740 spectra is useful to monitor transpiration over a long period such as day-long and season-long.Finally, the linkage of R740/R800 and DCI to PRI and WBI suggests the possible use of these indices in ecosystem scale and flux studies [36][37][38].

CONCLUSIONS
The experimental results presented here demonstrate sun-light induced chlorophyll fluorescence emitted from PSII and PSI are observable using reflectance indices associated with fluorescence peak wavelengths (690 and 740 nm) including fluorescence ratio index (FRI; R690/ R600 and R740/R800), curvature index (R675 × R690)/ R683 2 ) and derivative chlorophyll indices (DCI; D705/ D722, D730/D706 and (D720 − D703)/D707).Measurement data were collected four times using the crown of a mature naturally grown valley oak (Q.lobata) in the field over a one month period.During the experiment, solar irradiance was found to be the super-saturation condition for carbon assimilation of Q. lobata leaves.Changes in solar irradiance induced variations in PSII chlorophyll fluorescence parameters such as v m F F   and qP.Theoretically, strong light stress caused enhanced formation of damaging singlet oxygen in PSII [16] that degrade the D1 protein [26], resulting in decreases in PSII fluorescence.This phenomenon was observed at both leaf and crown scale using indices associated with F690 spectrum (r 2 = 0.35) At the same time, most of absorbed photon energy was considered to be dissipated as heat energy.Increases in leaf temperature decrease Rubisco activity inducing increases in the Ci/Ca ratio, accumulation of NADPH and reduction of PQ molecule.This physiology contributes to increases in PSI fluorescence emission [28].Results of crown scale measurements revealed variations in PSI fluorescence can be observed using reflectance indices associated with F740 (r 2 = 0.360 to 0.377).Also, the same reflectance indices were found to be useful in estimating stomatal conductance (r 2 = 0.494 to 0.511) and transpiration (r 2 = 0.537 to 0.662) because both F740 spectrum and these physiological responses were under the influence of leaf temperature mediated through Xan-thophyll cycle, not change in solar irradiance quality.
The above results demonstrate the potential use of reflectance indices based on chlorophyll fluorescence peaks for crown scale detections of plant physiological responses to natural environment, especially solar irradiance.The concluding remark of this paper suggests that crown scale transpiration can be monitored through remote estimation of PSI fluorescence over the season and can be expanded from crown scale to landscape scale according to the strong linkages to PRI and WBI.In addition, the results of this paper will contribute to chlorophyll fluorescence measurement from satellite platform for the use of estimating large scale transpiration.

Figure 1 .
Figure 1.Experimental site at big Chico Creek Ecological Reserve, CA, USA with micrometeorological station, study tower, study tree Quercus lobata (valley oak).
), leaf temperature (˚C) and PAR (µmole photon s −1 •m −2 ).The chlorophyll fluorescence parameters of steadystate fluorescence (F s ), quantum yield yield  v m  F F   [21] (Maxwell and Johnson, 2000) were measured along with photochemical quenching (qP), and the fraction of the allocation of absorbed light energy by PSII antenna; 1 ,

Figure 2 .Figure 3 .
Figure 2. (a) Time course measurement of the solar irradiance.Global radiation (W•m −2 ) was measured on September 29 th , October 6 th , 7 th , 28 th and November 4 th .The data point on the graph starts at 10:30 on September 29 th and moves to the next data at 15 minutes interval until it reaches 16:30.The other 4 measurement days repeats same graphing pattern.A daily cycle completes with 25 measurement data point.Therefore, September 29 th completes its diurnal cycle from Time course number to 25.Time course number 26 to 50, 51 to 75, 76 to 100 and 101 to 125 represents October 6 th , 7 th , 28 th and November th ; (b) Air temperature (˚C); (c) Relative humidity (%); (d) Leaf temperature (˚C).

Figure 5
show regression analysis of PAR with light use efficiency (LUE), chlorophyll fluorescence parameters and the allocation of absorbed light energy by PSII antenna.LUE and chlorophyll fluorescence parameters ( v m F F   , ΦPSII and qP) had a significant negative correlation with PAR (r 2 = 0.561 for LUE, 0.648 for v m F F   , 0.730 for ΦPSII and 0.699 for qP).The allocation of absorbed light energy by PSII antenna showed different responses (Figure 5(c)).Heat dissipation indicated by 1 − v m F F  hada positive correlation with PAR (r 2 = 0.648).
F s decreased from 20˚C to 28˚C and then stopped

Figure 5 .
Figure 5. (a) Regression analysis between PAR and LUE.All data have P-value < 0.001 (n = 78).(b) Regression analysis of PAR with chlorophyll fluorescence parameters ( v m F F   , ФPSII and qP) and (c) the fraction of the allocation of absorbed light energy by PSII antenna. 1 ,

Figure 6 .
Figure 6.Daily mean spectra of crown reflectance and crownderivative.Daily mean spectra were calculated with data taken from 10:30 to 16:30 each day.

Figure 8 .
Figure 8. Regression analysis of a derivative chlorophyll index, R690/R600, with (a) PAR; (b) chlorophyll fluorescence parameters; (c) the fraction of the allocation of absorbed light energy by PSII antenna and (d) the estimated rate of light energy allocation.All data have P-value < 0.001 (n = 78).
The estimated rate of light energy allocations (in terms of PAR) by PSII was successively tracked by R690/R600 (r 2 = 0.595 for PAR ×

Table 1 .
Summary of regression analyses between remote sensing and leaf level measurements.