Stress-induced changes in photosynthesis and proximal fluorescence emission of turfgrass

Remote measurements of solar-induced fluorescence (SIF) are now widely used to model gross primary productivity (GPP). However, the ability of SIF to track GPP in different environments, conditions, and at different scales remains uncertain. We designed an experiment to measure SIF and photosynthesis while inducing plant stress in replicated turfgrass. Immediately following application of abscisic acid (ABA), treated grasses experienced a 75% decrease in photosynthesis and an 18% decline in SIFyield, with evidence of alterations in energy partitioning. Withholding water resulted in slower photosynthetic inhibition of lower magnitude, with full recovery upon rewatering. In both treatments, reductions in SIF co-occurred with reductions in canopy greenness. However, we did not observe a relationship between the SIF and near-infrared reflectance of vegetation (NIRv) responses to our treatments in turfgrass. The response differences between treatments highlight the nuances of using SIF as a proxy for GPP.


Introduction
Uncertainties in future climate projections arise in part from our incomplete understanding of photosynthesis and ecosystem responses to stress.Sensors that measure photosynthesis at the scales of individual leaves are common and well characterized, however, it is difficult to scale these measurements to the whole plant, let alone to the stand, region, or globe (Ryu et al 2019).While there is no direct way to measure carbon uptake using spacebased platforms, Solar-Induced Fluorescence (SIF) emitted by illuminated chlorophyll molecules can be quantified by several current and planned satellites (Doughty et al 2022), including TROPOMI (Guanter et al 2021), OCO-2 (Sun et al 2017), OCO-3 (Taylor et al 2020), and GOSAT (Joiner et al 2011).Satellite-based SIF measurements have been shown to improve models of the global carbon cycle because of their positive, linear relationship with estimates of gross primary productivity (GPP) -the total photosynthetic carbon uptake at the ecosystem scale (Guanter et al 2014, Mohammed et al 2019, Schimel et al 2019).However, uncertainty remains in the consistency of the SIF-photosynthesis relationship across scales, taxa, and environmental conditions (Li et al 2020).Tower-and drone-based observations of SIF, so-called proximal sensing, coupled to physiological measurements offer an important means to assess photosynthetic properties and stress responses of vegetated plots or individual tree canopies (Grossmann et al 2018, Porcar-Castell et al 2014, 2021).
Photosynthetically active radiation (PAR) drives both photosynthesis and SIF (Parazoo et al 2020).The actual SIF measured from above the canopy is a product of PAR, the fraction of PAR absorbed by the leaf (f PAR ), the quantum efficiency of fluorescence ( F f ), and the probability that a fluoresced photon escapes the canopy without being absorbed (f esc ) (Dechant et al 2020).
he product of PAR and f PAR is often referred to as absorbed photosynthetically active radiation (APAR).The parallel dependencies SIF and photosynthesis on PAR are complicated by dynamic partitioning of APAR by plants among: (1) photochemical light use in support of linear electron transport, (2) non-photochemical quenching (NPQ) through thermal energy dissipation, and (3) SIF (Porcar-Castell et al 2021).Further, differences in instrument design and operation can affect retrieved values of SIF, resulting in reports varying by more than an order of magnitude (Marrs et al 2021).This makes intercomparison across experiments and field studies challenging.
Experimental manipulations intended to impose stress and/or inhibit photosynthesis have yielded variable results in studies of proximally measured SIF.Foliar treatment with the herbicide DCMU (3-(3,4dichlorophenyl)−1,1-dimethylurea, which blocks re-oxidation of photosystem II (PSII) thereby inhibiting photosynthetic electron transport, led to an increases in SIF in turfgrass sod (Rossini et al 2015), maize, and wheat (Pinto et al 2016) .In these studies, increasing stress was associated with increasing SIF, presumably as the inhibition of photochemistry led a greater fraction of absorbed energy to be partitioned to fluorescence.In contrast, water-stressed plots of potato exhibited lower SIF when compared with irrigated plots (Xu et al 2021).A sensitivity analysis of this finding revealed that reduced SIF resulted from two primary factors: (a) steeper leaf angles induced by drought (i.e., wilting), reducing APAR and the probability that fluoresced photons would escape the canopy and reach the sensor, and (b) increased levels of NPQ in response to drought, which consumed a greater fraction of absorbed energy, resulting in lower partitioning to fluorescence (Xu et al 2021).Some studies fail to observe a significant response of SIF to photosynthetic inhibition (Helm et al 2020).For example, Marrs et al (2020) inhibited photosynthesis of two species of temperate deciduous broadleaf trees by driving stomatal closure via either foliar application of abscisic acid (ABA) or the introduction of xylem emboli.In response to both treatments, light-saturated rates of photosynthesis fell by more than 80% while measured SIF did not change.
We designed turfgrass manipulations to assess the effects of stress on photosynthesis, SIF, and an array of widely used remote sensing indices of productivity.Turfgrass was chosen because managed lawns are a significant fraction of urban and sub-urban landscapes, and because stress treatments could be precisely imposed on grasses and monitored.We subjected turfgrass to either drought or foliar application of ABA.These treatments were chosen because each inhibits photosynthetic CO 2 assimilation through stomatal closure, however, previous observations suggest that their effects on energy partitioning may differ in magnitude.Drought has been shown to lead to increased levels of thermal energy dissipation correlated with decreases in photochemical light use (Helm et al 2020, Xu et al 2021), whereas a previous report of the effects of foliar ABA treatment on temperate trees suggested an increase in thermal energy dissipation of a relatively smaller magnitude (Marrs et al 2021).We hypothesized that SIF should more significantly track photosynthesis metrics than the vegetation indices NDVI and NIRv, particularly at the short time scales of our study.We further hypothesize that correlations between SIF and photosynthesis will be strongest when stress leads to enhanced engagement of thermal energy dissipation.Simultaneous measurements of SIF and spectral reflectance were paired with leaf-level measurements of gas exchange and pulse-amplitude modulated (PAM) chlorophyll fluorescence emission.Replicated pallets of carefully managed turfgrass were subjected to either foliar ABA treatment or low soil water availability (by withholding water).Both treatments affect stomatal closure and photosynthesis, but with different timing and magnitudes.

Grass preparation
Nine weeks before our study, we sowed Top Choice Tall Fescue seed (M6M-21-TCTF1, Newsom Seed, Inc. Fulton, MD, USA * , Festuca arundinacea Schreb.), a seed source commonly used by turfgrass farmers in the vicinity of our field site, at high density on the surface of MetroMix 360 (Sun Gro Horticulture, Agawam, MA, USA) lined plastic pallets measuring approximately 70 cm x 60 cm filled with soil to a depth of 5 cm.Pallets were maintained in [a greenhouse] until transport to the field site two days prior to the initiation of our study.The greenhouse is full-sun exposed and maintained between 15 and 36 °C (20.8 °C ± 3.3 °C, mean ± 1 s.d.).Pallets were well-watered and fertilized weekly starting one week after leaf emergence.The grass canopy was clipped by hand weekly to maintain uniform height.We varied the positions of pallets daily to minimize any effects of heterogeneous greenhouse microenvironments.

Experimental design
We conducted our study at the Forested Optical Reference for the Evaluation of Sensor Technology (FOREST) field site on the campus of the National Institute of Standards and Technology in Gaithersburg, Maryland, USA (Winbourne et al 2022).The grass pallets were placed in an open field which received direct sunlight from morning until evening each day.SIF and other vegetation indices were measured using a system of two spectrometers.These spectrometers were housed in a custom temperature-controlled enclosure and fibercoupled to a 2-inch aperture telescope.The specifications for the spectrometers and fiber were based on the PhotoSpec instrument (Grossmann et al 2018).Although our system, like PhotoSpec, was designed on a twoaxis rotational platform, we chose to keep the optics fixed for the duration of this experiment to limit variability due to sun-sensor geometry.The telescope was fixed to a rigid mount on the roof of a building 11.2 m above the ground and directed downward at an angle of 37.4°.The targeted ground area, 18.4 m away, was located at night by back-filling the fore-optics using a fiber-coupled halogen bulb.Shining a light through this system also allowed us to see the field-of-view of the instrument, which was an elliptical spot 19 cm wide and 27 cm long on the ground.This area was marked and a mount was fixed on the ground to consistently align the center of the measurement pallet and maintain constant geometry with the optics across all measurements.The entire instrument, including the extension fiber and fore-optics were radiometrically calibrated using an integrating sphere.This sphere was in turn calibrated using a NIST-traceable spectroradiometer.The experimental design is illustrated in figure 1.
During periods of clear sky conditions, pallets were placed into the field of the view of the sensor for a period of 30 s.After this period, 30 s was allowed for the removal of this pallet and the placement of the next pallet, allowing us to measure a new sample every minute.After every three pallets, a polytetrafluoroethylene (PTFE) based reflectance panel reference target was placed in the field of view for 30 s. Measurements of the reference panel were used to compute the downwelling radiation spectra needed for the data retrievals.The entire measurement cycle took 20 min, including the 15 pallets and five observations of the reference.Sky conditions were carefully monitored and measurement cycles were paused as needed to allow any clouds to pass.Downwelling PAR was also continuously measured next to the pallet mount using a Hobo PAR sensor and Micro Station Data Logger (Onset Computer Corporation, Bourne, MA, USA).The experiment lasted four days, with proximal measurement cycles performed one or two times each day, depending on sky conditions.After the imposition of treatments, grass canopy temperature was measured using a handheld infrared thermometer.Each day, prior to any measurements, a metal rake was used to gently orient the tips of the grass canopy perpendicular to telescope's line of sight.This was done to minimize variations in observed SIF between pallets due to leaf-angle effects.

ABA and water stress treatments
At day 0, all 15 grass pallets in the study were well watered and uniform.The pallets were randomly divided into 3 treatment groups with 5 pallets each.We inhibited photosynthesis in two of the groups by inducing stomatal closure via two treatments, foliar application of ABA or withholding water.In the first group, the canopy (both leaf surfaces) was sprayed (using a handheld spray bottle) with 0.1 mM ABA in 0.1% Triton X-100, made up as in Marrs et al (2020), seven times between 07:30 and 08:00 on the morning of what is hereafter referred to as day 1.The canopy had dried before measurements of any kind were made that day.In the second group, water was withheld to impose physiologic stress associated with low soil water availability.Under the warm and sunny summer conditions in Maryland, USA, the thin soils (5 cm) of each pallet dried quickly.Gravimetric soil water content (GSWC, mass of water per mass of dry soil) of moisture-stressed pallets fell from 1.71 to 0.68 by the end of day 1, the first day water was withheld, and fell further to 0.63 by the end of day 2. At the end of day 2, to avoid widespread grass mortality, the water-stress treatment pallets were watered to their well-watered pallet mass by distributing water evenly across each pallet.GSWC returned to near pre-treatment values (1.5) after watering.GSWC of control and ABA-treated racks never fell below 1.2 and was generally above 1.6.The third set of 5 pallets, the control group, were not subjected to ABA spraying or water stress.Control and ABA-treated pallets were watered up to their well-watered mass each afternoon, after measurements were made.

Leaf-Level measurements
Leaf-level photosynthetic CO 2 assimilation and pulse-amplitude modulated (PAM) chlorophyll fluorescence emission were measured through the middle of each day using a Licor LI-6800 photosynthesis analyzer with the 6800-01 A fluorometer head (LI-COR Biosciences, Lincoln, NE, USA).To perform measurements, three grass blades from within the focal area of the spectrometer were isolated by feeding them gently through a plastic mesh.The width of each blade was determined before they were centered in the chamber of the analyzer adjacent to each other.Leaf-level measurements were made using an internal lamp set to 1600 μmol m −2 s −1 and a reference CO 2 concentration of 400 μ mol mol −1 .The internal block temperature was set slightly above midday ambient temperature for each particular day to minimize condensation in the instrument (36 °C for day 0 and day 1, 35 °C for day 2, and 32 °C for day 3), and the internal relative humidity was set close to midday conditions on each day (yielding vapor pressure deficits of 3.40 ± 0.06, 3.10 ± 0.06, 3.30 ± 0.04, and 2.60 ± 0.02 kPa for days 0-3, respectively.)The actual quantum yield of photosystem II photochemistry during illumination (f PSII ) was calculated as ( .Levels of energy dissipation can also be quantified as NPQ, however, measuring NPQ is impractical in most field settings due to its dependence on dark-acclimated F m (Logan et al 2007, Maxwell andJohnson 2000), and was not performed in this study.

Calculation of vegetation Indices and SIF
The Ocean Insight Flame spectrometer records spectra from 340 to 1045 nm, with 0.38 nm resolution.Spectra from this instrument were used to quantify three common vegetation indices (VIs).The reflectance at a particular wavelength λ was computed as the ratio of radiance from the grass rack at that wavelength (E λ ) to the radiance from the white reference target closest in time (L λ ).
The normalized differential vegetation index (Carlson andRipley 1997, Tucker 1979) was calculated as the dominance of near-IR reflectance compared to red reflectance as shown in equation (3).For the calculation of SIF, several retrieval methods exist, each with different assumptions and uncertainties (Chang et al 2020).The Fraunhofer Line Depth (FLD) method (Plascyk 1975) was used in this experiment to be consistent with prior work done studying the effect of ABA on photosynthesis (Marrs et al 2020).This method exploits the atmospheric absorption around 760 nm due to the O 2 A-band.Grass radiance values (E) and radiance values taken from the white reference target (L) from within the O2-Aband (760.5 nm) and on the shoulder of the band (757.5 nm) were used.A high-resolution grating spectrometer (Ocean Insight QE-Pro) was used for the SIF retrievals.The spectrometer recorded spectra from 650.5 to 878.4 nm at 0.25 nm resolution.

= ´--
The SIF values obtained were then corrected for atmospheric absorption between the canopy and sensor (Sabater et al 2018) using pressure, temperature, and humidity data from the co-located weather station.SIF values were also calculated using a differential optical absorption spectroscopy (DOAS) retrieval method (Grossmann et al 2018).The values retrieved using the two methods were consistent (figures S1-S2).
Since the leaf-level measurements were made using a fixed-magnitude light source, SIF values were normalized to the magnitude of incoming light (PAR) for comparison.Spaceborne SIF sensors, with coarse spatial resolution, commonly have vegetated and non-vegetated areas within a single pixel.It is therefore common to also normalize SIF observations by NDVI, to calculate a SIF yield (equation ( 7)) of the vegetated portion of the field of view (Li et al 2020).This normalization is also useful for proximal measurements of SIF.Even though the field-of-view of the sensor is limited to the center of the grass targets, gaps between blades, canopy self-shading, and non-green portions of leaves can lead to lower SIF values when compared to leaf-level measurements.

Determination of treatment effects
In order to determine what effect the two treatments had on SIF and other parameters, data from treated racks was compared to the controls racks on each day of the experiment.The percent difference for each parameter was calculated using equation (8).percentile bootstrap analysis (Diciccio and Romano 1988) was performed to generate a 95% confidence interval on the magnitude of these effects.To determine the significance of the effects, a two-sample t-test (Keselman et al 2004) was performed as well.

Results
Foliar application of abscisic acid led to (>80%) inhibition of stomatal conductance which was observed on day 1 of treatment and sustained through the conclusion of the experiment on day 3 (figure 2(a)).Consistent with the reduction in transpirational cooling brought on by stomatal closure, the temperature of the canopy of ABAtreated pallets was higher than that of control pallets on each of the three days after treatment (figure S3).Stomatal closure was also associated with the inhibition of more than 70% of CO 2 assimilation (figure 2 2(c)).Both effects persisted throughout the remainder of the experiment.On the day of ABA application, mean solar-induced fluorescence intensities of treated grasses were 18% below those of controls, and both SIF and SIF yield remained suppressed over the ensuing two days (figures 2(g), (h)).The NDVI and NIRv of ABA-treated grasses declined relative to untreated grasses on day 2 and fell further on day 3 (figures 2(d)-(e)).This decline in NDVI was associated with a perceptible 'yellowing' of ABA-treated grasses (personal observation) and with a lower Green Chromatic Coordinate (GCC) (Richardson et al 2007) calculated from color intensities of digital images of grass canopies (figure S4).PRI similarly declined on days 2 and 3 after treatment (figure 2(f)).
Low soil water availability resulting from withholding water drove declining stomatal conductance and was alleviated on day 3 after rewatering (figure 2(a)).There was also an increase in canopy temperature observed on day 3 in the water-stressed grasses (figure S3).Similarly, photosynthetic CO 2 assimilation (figure 2(b)) declined in response to low soil water availability, falling by 55% when compared to the control pallets on day 2, and recovering after rewatering.In contrast with foliar ABA treatment, and contrary to expectations, the imposition of low soil water availability had no significant effect on F F v m / ¢ ¢ (figure 2(c)).Solar-induced fluorescence intensities and SIF yield of water-stressed grass were significantly lower than those of untreated grass on day 3 only, the day after rewatering.Also on day 3, NDVI of water stressed grasses fell (figure 2(e)), along with a decrease in GCC and a visibly bluer coloration (personal observation).In contrast to the ABA treatment, the NDVI and NIR v were decoupled, with NIRv increasing immediately with low soil water and then showed no difference from the control after rewatering.PRI differed significantly from control values only on day 3, after rewatering.
A linear relationship between A net and f PSII persisted throughout the course of the experiment and across the treatments (figure 3(a)).The PRI relationship with A net shifted with ABA treatment on days 2 and 3. PRI, SIF, and SIF yield did not show strong, consistent relationships with A net (figures 3(b)-(d)).

Discussion
Foliar ABA treatment and the imposition of low soil water availability both resulted in stomatal closure leading to photosynthetic inhibition in turfgrass, albeit with differing timing and magnitudes.Both treatments also led to reductions in SIF.In contrast with satellite-based studies (Badgley et

ABA treatment
Foliar ABA treatment resulted in rapid stomatal closure leading to decreased carbon uptake efficiency in turfgrass, and this effect lasted for the duration of the experiment (figures 2(a)-(b)).This decrease in carbon uptake efficiency was also detected by the PAM fluorescence measurements (figure 3(a)).ABA-treated palettes also saw day one decreases in SIF and SIF yield (figures 2(g), (h)), coincident with the decreases in intrinsic Carotenoids of the xanthophyll cycle serve as precursors to endogenous ABA biosynthesis.ABA treatment can result in the accumulation of carotenoids, chiefly zeaxanthin, and the induction of energy dissipation (Barickman et al 2014).Foliar zeaxanthin accumulation can be observed non-destructively through its effects on PRI, and lower values of PRI result from zeaxanthin accumulation and from an increase in the ratio of carotenoids to chlorophylls (Gamon et al 1992).By day 2 after the ABA treatment, PRI was lower when compared to untreated turfgrass and decreased further by day 3. ABA is a stress hormone that not only signals stomatal closure, it can also lead to chlorophyll degradation.Chlorophyll degradation, especially when combined with carotenoid accumulation, would affect leaf coloration (i.e., result in observed leaf yellowing) and explain the decline in the greenness index, NDVI, we observed from day 1 to day 3.A decrease in leaf chlorophyll content could also explain the further declines in SIF from day 1 to day 3.Our derivation of SIF yield accounted for treatment effects on NDVI.
Our observations suggest that ABA-treated leaves adapted to the stress, in part, by increasing photoprotective energy dissipation (non-photochemical quenching, NPQ), as detected by decreases in the PAM fluorescence measurement F F v m / ¢ ¢ and by the decrease in PRI.

Withholding water treatment
Stress was also induced by limiting soil water availability.While this stress also induced stomatal closure and reduced carbon uptake, the effect was not as severe or immediate when compared to the effect of ABA application (figures 2(a), (b)).Likewise, effects on F F v m / ¢ ¢ and PRI were subtle and mostly not statistically significant, although, based upon prior reports of broad-leafed plant species (Helm et al 2020, Xu et al 2021) we hypothesized that we would observe greater engagement of energy dissipation in response to drought when compared to foliar ABA treatment.This may be because stomatal closure results in elevated rates of photorespiration in C3 plants such as tall fescue, as CO 2 concentrations fall in the leaf interior.Like photosynthesis, the photorespiratory pathway requires reductant.This requirement can be met by photochemistry, which, in combination with non-assimilatory electron flow to oxygen (Osmond and Grace 1995) buffers against large responses in the rate of photosynthetic electron transport as the balance of photosynthesis to photorespiration shifts during water stress.On day 3, after rewatering, stomatal conductance and photosynthetic CO 2 assimilation recovered fully from prior low soil moisture availability.However, it was only on day 3 that SIF declined significantly in this treatment.This co-occurred with decreases in NDVI and GCC.Changes in leaf coloration could have resulted from the accumulation of anthocyanins and exerted an effect on SIF by reducing PAR absorbed by chlorophyll (Gould 2004).Interestingly, in the water-stress treatment, our derivation of SIF yield did not fully compensate for the effects of lower NDVI on SIF.These changes in leaf coloration differed in nature from those induced by ABA treatment and warrant further examination.

Conclusion
Three prevailing factors affect SIF intensity: APAR, the quantum yield of fluorescence, and the probability of fluorescence escape towards the sensor (Dechant et al 2020).Both of the treatments we imposed on turfgrassfoliar ABA application and low soil water availability-led to changes in leaf coloration.Although these changes in coloration differed in their timing, extent, and magnitude, each could have reduced PAR absorbed by chlorophyll, thereby explaining co-occurring reductions in SIF.Contrary to our expectations, only ABA treatment led to increased levels of thermal energy disspation.However, our findings are consistent with our prediction that correlations between SIF and photosynthesis would be strongest when photosynthetic inhibition was accompanied by enhanced engagement of thermal energy dissipation.Treatment with ABA resulted in reductions in SIF that co-occurred with the induction of energy dissipation, an alternative energy partitioning pathway in photosystem II that would reduce fluorescence quantum yield.Thus, alterations to key factors influencing SIF intensity can explain SIF dynamics in response to both treatments.Taken together, our data suggest that when there is evidence for further engagement of thermal energy dissipation (NPQ), SIF better tracks the timing but not the magnitude stress-induced declines in photosynthesis.In the absence of stressinduced changes to NPQ, SIF fails to track both the timing and magnitude stress-related photosynthetic declines in our study.
More so than any other remotely measured product, SIF yield was able to able to detect declines in photosynthesis on the first day of ABA treatment, and for several days afterwards.Other remote products (NDVI, NIRv, and PRI) also showed changes which can be attributed to stress, but these changes were not seen until one day after treatment and were not as consistent.The suite of vegetation indices detected changes in plant structure, pigment content, and NPQ, showing the utility of collocating these measurements with observations of SIF.However, SIF failed to detect the slower decline in photosynthesis observed in experimental drought and the photosynthetic recovery in the experimentally droughted turfgrasses upon rewatering.This suggests the need for caution in the use of SIF as a proxy for photosynthesis at short temporal scales.Further, our results also suggest that care must be taken in interpreting the magnitude of change in SIF relative to changes in photosynthesis.

Disclaimer
Certain commercial equipment, instruments, or materials are identified in this paper in order to specify the experimental procedure adequately.Such identification is not intended to imply recommendation or endorsement, nor is it intended to imply that the materials or equipment identified are necessarily the best available for the purpose.

Open research
The data presented in this work can be found in the following repository: https://dataverse.harvard.edu/dataverse/bu_turfgrass_experiment/.

Figure 1 .
Figure1.Diagram of the field experiment configuration.The observation elevation angle, θ OEA , was constant throughout the experiment.The relative azimuth angle θ AZ and solar elevation angle θ SEA were dependent on time of day, however each full round of spectral observations was performed within approximately 20 min, rendering between-pallet differences negligible.All pallets were kept in the field between measurements under uniform conditions and cycled into the fixed target mount for measurement.

(
Maxwell and Johnson 2000), where F m ¢ is maximal fluorescence emission during a 0.8 s saturating pulse of light and F s is steady-state fluorescence emission during illumination.The intrinsic quantum yield of photosystem II photochemistry during illumination was calculated as F F -¢ Here F o ¢ is the minimal fluorescence emission measured upon the transient removal of actinic light in the presence of weak far-red illumination imposed to facilitate complete oxidation of PSII reaction centers.Energy dissipation involves the safe conversation of excitation energy to heat.Induction of energy dissipation therefore results in decreases in the intrinsic quantum yield of photochemistry, measurable as decreases in F F v m / ¢ ¢(Logan et al 2007, Marrs et al 2020) photochemical reflective index (PRI)(Gamon et al 1997) was designed to discern the relative abundance of pigments associated with the xanthophyll cycle.( ) PRI 5 531 570 531 570 r r r r = -+ al 2017, Dechant et al 2022, Mengistu et al 2020), we did not observe consistency between the SIF and NIRv responses to stress in turfgrass, with NIRv being decoupled from F F v m / ¢ ¢ (figures 2(c), (e)) and showing opposite sign responses between the two stress treatments.NIRv very closely tracked the patterns in NDVI and PRI.While all three of these vegetation indices were able to detect plant stress after a few days, only SIF and SIF yield showed significant change on the first day following ABA treatment.

Figure 2 .
Figure 2. Mean differences between control and treated palettes across the four days of the experiment.Error bars represent 95% confidence intervals.Approximate timing of the application of abscisic acid (ABA) and rewatering of the drought stressed racks are shown in red and blue dashed lines respectively.p-value levels indicating the significance of the treatment effects are shown with stars ( * p < 0.05, ** p < 0.01, *** p < 0.001).

Figure 3 .
Figure 3. Mean values for each palette on each day of net carbon assimilation (A net ) compared to (a) the quantum efficiency of photosystem II, (b) the photochemical reflective index, (c) solar-induced fluorescence, and (d) SIF yield .