Low-Cost Chlorophyll Fluorescence Imaging for Stress Detection

Preparing to load PDF file. please wait...

0 of 0
100%
Low-Cost Chlorophyll Fluorescence Imaging for Stress Detection

Transcript Of Low-Cost Chlorophyll Fluorescence Imaging for Stress Detection

sensors

Article
Low-Cost Chlorophyll Fluorescence Imaging for Stress Detection
Reeve Legendre 1 , Nicholas T. Basinger 2 and Marc W. van Iersel 1,*

1 Department of Horticulture, University of Georgia, Athens, GA 30602, USA; [email protected] 2 Department of Crop & Soil Sciences, University of Georgia, Athens, GA 30602, USA;
[email protected] * Correspondence: [email protected]

Citation: Legendre, R.; Basinger, N.T.; van Iersel, M.W. Low-Cost Chlorophyll Fluorescence Imaging for Stress Detection. Sensors 2021, 21, 2055. https://doi.org/10.3390/ s21062055
Academic Editor: Akio Kuroda
Received: 20 February 2021 Accepted: 13 March 2021 Published: 15 March 2021
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

Abstract: Plants naturally contain high levels of the stress-responsive fluorophore chlorophyll. Chlorophyll fluorescence imaging (CFI) is a powerful tool to measure photosynthetic efficiency in plants and provides the ability to detect damage from a range of biotic and abiotic stresses before visible symptoms occur. However, most CFI systems are complex, expensive systems that use pulse amplitude modulation (PAM) fluorometry. Here, we test a simple CFI system, that does not require PAM fluorometry, but instead simply images fluorescence emitted by plants. We used this technique to visualize stress induced by the photosystem II-inhibitory herbicide atrazine. After applying atrazine as a soil drench, CFI and color images were taken at 15-minute intervals, alongside measurements from a PAM fluorometer and a leaf reflectometer. Pixel intensity of the CFI images was negatively correlated with the quantum yield of photosystem II (ΦPSII) (p < 0.0001) and positively correlated with the measured reflectance in the spectral region of chlorophyll fluorescence emissions (p < 0.0001). A fluorescence-based stress index was developed using the reflectometer measurements based on wavelengths with the highest (741.2 nm) and lowest variability (548.9 nm) in response to atrazine damage. This index was correlated with ΦPSII (p < 0.0001). Low-cost CFI imaging can detect herbicide-induced stress (and likely other stressors) before there is visual damage.
Keywords: chlorophyll fluorescence imaging; stress; herbicide; pixel intensity; PAM fluorometry; photosystem II
1. Introduction Plants naturally contain high levels of chlorophyll a, a fluorophore that is sensitive to a
wide range of environmental stresses. Chlorophyll a fluorescence was first viewed in 1931 by Kautsky and Hirsch [1], nearly 90 years ago, by using a red-transmitting filter to view a plant that was moved from darkness to blue light. Since then, chlorophyll fluorescence has been developed into a tool that can quantify photosynthetic efficiency and detect a variety of stresses that affect the photosynthetic apparatus. Chlorophyll fluorescence is directly related to the photosynthetic efficiency of plants, because there are three possible fates of the energy of photons that have been absorbed by photosynthetic pigments. The energy form the photons can be used to drive the light reactions of photosynthesis (electron transport), be converted into heat, or be re-emitted as fluorescence by chlorophyll a [2]. The quantum yields of these three processes add up to one, so if the quantum yield of one process decreases, the quantum yield of one or both other processes will increase. The variability in chlorophyll fluorescence can thus be used as a measure of photosynthetic efficiency and primarily relates to changes in the efficiency with which photosystem II (PSII) uses the excitation energy from absorbed photons [3], because nearly all variable chlorophyll fluorescence comes from chlorophyll a surrounding PSII [4]. One of the primary strengths of chlorophyll fluorescence measurements is that many stresses can be detected before any signs of visible damage occur [5–7]. This is dependent on the type of stress incurred by the plant and has the potential to help minimize crop production problems

Sensors 2021, 21, 2055. https://doi.org/10.3390/s21062055

https://www.mdpi.com/journal/sensors

Sensors 2021, 21, 2055

2 of 17
through early detection of stressors that may negatively impact the crop, followed by appropriate recourse.
Chlorophyll fluorescence measurements can detect the physiological effects of a wide variety of stressors, including abiotic stressors such as high light, extreme temperatures, or drought [8–13], and chemical stressors like herbicides or heavy metals [11,14–18]. There are other applications in plant disease detection and screening for disease resistance in a variety of plant species [9,19–22]. The most common method to measure chlorophyll fluorescence uses pulse amplitude modulation (PAM) fluorometry. PAM fluorometers allow for measurement of chlorophyll fluorescence under ambient light conditions by pulsing low intensities of a measuring light. By measuring the amount of fluorescence emitted with and without the small pulses, PAM fluorometry can quantify the fluorescence induced by the measuring pulse only. PAM fluorometry requires measurements under both ambient light and during a saturating light pulse to determine the quantum yield of PSII (ΦPSII), which is a measure of the operating efficiency of PSII [23,24]. The saturating light pulse needs to be intense enough to fully saturate all photosynthetic reaction centers and typically has an intensity of 6000 to 10,000 µmol m−2 s−1, approximately 3 to 4 times to maximum photosynthetic photon flux density plants ever are exposed to under natural conditions. The requirement for such an extremely intense light pulse creates limitations, since it is infeasible to apply such a pulse to a large crop area, while also unsafe for people who might be exposed to such intense light. An additional measurement of chlorophyll fluorescence in the dark, using the measurement light only, is necessary to quantify changes in non-photochemical quenching (NPQ). The NPQ parameter describes how much heat dissipation is upregulated compared to a dark-adapted leaf, when heat dissipation is minimal. In addition to ΦPSII and NPQ, chlorophyll fluorometry can measure other, less common, photochemistry-related parameters: the quantum yield of non-photochemical energy dissipation in response to light exposure (ΦNPQ), and the quantum yield of other, non-light induced energy dissipation processes (ΦNO).
Many PAM fluorometers use fiber-optic cables to send the measuring light to a leaf and to capture the resulting fluorescence. Such systems are limited to point measurements, which cannot account for spatial variability that may be present across a leaf or canopy. Accurate assessment of the photosynthetic performance of a plant or crop can require many measurements to ensure that those samples are representative of the entire canopy.
A more recent development in monitoring chlorophyll fluorescence is chlorophyll fluorescence imaging (CFI), which also uses the principles of PAM fluorometry. Chlorophyll fluorescence imaging allows for visualization and quantification of the photosynthetic heterogeneity of an entire leaf or small canopy at once [25]. The PAM measuring light can be provided by pulsed lasers or light-emitting diodes (LEDs). When paired with the CCD camera and a control unit, the camera can be synced with the pulsed light to capture the fluorescence resulting from the measuring light. Since the chlorophyll a fluorescence spectrum peaks at 690 and 740 nm [26], CFI systems typically use a camera with a long-pass filter, while blocking shorter wavelengths. Once the images are captured, they are segmented and fluorescence parameters are calculated for each pixel, followed by visualization to produce chlorophyll fluorescence images [25]. Such CFI systems have been used successfully to detect damage from phytotoxic compounds [27], including herbicides [28]. However, due to the requirement for a saturating light pulse, commercially-available CFI systems are limited to individual leaves or small plants. In addition, commercially-available CFI systems require specialized electronics for image acquisition and processing and are expensive, while they often have low resolution.
Due to the complexity and expense of current CFI systems, a less expensive alternative that does not require PAM fluorometry for early stress detection would be beneficial, especially if it allows for measurements at larger scales. While the PAM-based CFI systems provide highly detailed, quantitative measurements, capturing images that simply display chlorophyll fluorescence may provide enough information to detect stress before it would otherwise be visible.

Sensors 2021, 21, 2055

3 of 17
A simple, inexpensive CFI system can be assembled using a digital camera with a long pass filter, allowing red and/or far-red light to pass, a light secure enclosure, and blue light to drive photosynthesis [29]. There are limitations to such systems, as it is not possible to calculate the fluorescence parameters acquired using PAM fluorometers, like ΦPSII, NPQ, ΦNPQ or ΦNO [30]. However, a simple CFI system may be able to visualize plant stress. We conducted two studies to determine whether a simple imaging system is capable of visualizing changes in the two most common chlorophyll fluorescence parameters.
In the first study, we used a simple CFI system to capture chlorophyll fluorescence images of plants exposed to the herbicide atrazine to examine whether this system can be used for early stress detection, using pixel intensity as a direct measurement the fluorescence intensity. Atrazine (2-chloro-4-ethylamine-6-isopropyl-amino-s-triazine) is a triazine herbicide used primarily as a broad-spectrum control agent for broadleaf and grass weeds. Atrazine’s mode of action is as a PSII inhibitor that competitively binds to plastoquinone-binding proteins of PSII. Doing so prevents normal electron transport required for the light reactions of photosynthesis, which results in an inhibition of photosynthesis and extensive oxidative damage in chloroplasts [31–34]. The direct inhibition of PSII by atrazine makes it a good candidate for use in our study since inhibition of PSII activity, and thus photosynthetic electron transport, is expected to increase chlorophyll fluorescence. In addition to imaging the chlorophyll fluorescence, measurements of the leaf reflectance/fluorescence and ΦPSII were taken to look for correlations among these measurements. We hypothesized that the application of atrazine would result in an increased pixel intensity representative of damage to the photosynthetic apparatus of leaves and that this would be associated with a decrease in ΦPSII and an increase in measured leaf reflectance/fluorescence in the waveband where chlorophyll fluorescence emissions occur. In the second study, we determined whether our CFI system can visualize downregulation of NPQ after a plant that was exposed to relatively high light conditions is transferred to darkness. This typically results in downregulation of NPQ [35–37] and we therefore hypothesized that this would be accompanied by an increase in chlorophyll fluorescence.
2. Materials and Methods 2.1. Plant Materials and Herbicide Application
Three ‘Cora Punch’ vinca (Catharanthus roseus; Syngenta, Basel, Switzerland) plants in 10-cm pots filled with a peat-based substrate were purchased from a local garden center and used during this study. Two ‘Green Towers’ lettuce (Lactuca sativa) plants and two new guinea impatiens (Impatiens hawkeri) were also used, but results from those trials are not discussed in detail, since they responded essentially the same as the vinca plants. To induce stress, 250 mL of atrazine solution (AAtrex® 4L, Syngenta; 1.33 mg of active ingredient/L) was applied as a soil drench resulting in an application rate equivalent to field rates (2.25 kg of active ingredient/ha). Atrazine is highly mobile in soil or peat-based substrates, allowing the root system of the plants to take up the atrazine, followed by distribution throughout the plant.
For the duration of each run, one plant was placed inside a multispectral digital imaging system (Topview, Aris, Eindhoven, The Netherlands) and not moved after herbicide application. A cool-white LED panel (Cool white 225 LED ultrathin grow light panel, Yescom USA, City of Industry, CA, USA) was hung approximately 20 cm above the plant canopy and was used to drive photosynthesis and transpiration to promote herbicide uptake, movement, and physiological injury. The LED panel provided a photosynthetic photon flux density of approximately 175 µmol m−2 s−1 at the top of the canopy. Note that photosynthesis is a quantum-driven process and that light intensities therefore are typically reported as photon flux densities, rather than energy flux densities.
2.2. Reflectance/Fluorescence and ΦPSII
The fiber-optic cable of a spectrometer (Unispec Spectral Analysis System, PP Systems, Amesbury, MA, USA) was pointed at a leaf near the top of the canopy using a leaf clip to

Sensors 2021, 21, 2055

4 of 17
take leaf reflectance measurements. Reliable data could be detected at wavelengths from 450 to 770 nm due to the spectrum of the halogen bulb used by the spectrometer. Note that these leaf reflectance measurements in reality are a combination of the reflectance and chlorophyll fluorescence emitted from the leaves. The reflectance measurement immediately after herbicide application, before translocation of the atrazine to the leaves had occurred, was used as the baseline to normalize subsequent measurements. Since we were mainly interested in changes in chlorophyll fluorescence in response to the atrazine application, the normalized reflectance/fluorescence was averaged across the waveband of chlorophyll fluorescence (669.8–760.7 nm) to calculate the normalized average reflectance in the fluorescence spectrum (nARFS). The nARFS represents the change in the fluorescence spectrum from the initial timepoint.
A PAM fluorometer (Junior-PAM, Walz, Effeltrich, Germany) was used to take measurements of ΦPSII using a fiber optic cable aimed as closely as possible (<5 mm) at the location where the reflectance measurements were taken. Measurements using both the spectrometer and the PAM fluorometer were taken immediately prior to chlorophyll fluorescence imaging.
2.3. Chlorophyll Fluorescence Imaging
The cool-white LED panel was removed for ~1 min to facilitate digital imaging. For the chlorophyll fluorescence images, three plants (one plant per replication) were illuminated using blue LEDs, and a bandpass filter (650–740 nm) allowed only fluorescence to be captured by the monochrome digital camera in the Topview imaging system (Aris). Images were captured beginning immediately after herbicide application and subsequently every 15 min for 8 h. The monochrome images have an 8-bit resolution, resulting in a pixel intensity scale that spans from 0–255, where 0 represents a black pixel while 255 represents a white pixel, with varying shades of gray in between. Composite RGB images were taken at the same time using the multispectral imaging system, to compare to the fluorescence images and to see if symptoms of stress were visible in those RGB images (Supplementary Figure S1). Each of the three replications were done on separate days.
2.4. Image and Data Analysis
To quantify the pixel intensity, chlorophyll fluorescence images were analyzed using the Fiji software package (www.fiji.sc, accessed on 14 March 2021). A 50 × 50 pixels square area, as close as possible to the area where the reflectance and ΦPSII measurements were taken, was selected and the average pixel intensity calculated. For statistical analysis, all data was analyzed using JMP Pro (version 15.0.0) using generalized linear models between each factor to check for correlations between nARFS, ΦPSII, and pixel intensity. Since our approach is more qualitative than quantitative, each replication was analyzed separately.
The standard deviation of the normalized average reflectance was calculated for all measured wavelengths to determine which wavelengths were best suited for the development of a fluorescence-based stress index (FBSI). Using a similar approach as already in use for other plant- or crop-based indices can facilitate the development of a sensor to detect stresses with a chlorophyll fluorescence-based stress signature.
2.5. Pixel Intensity and Heat Dissipation
A petunia (Petunia × hybrida) plant was moved from a greenhouse into the imaging system, where it was allowed to fully dark-adapt for 20 min, resulting in the opening of all photosystem II reaction centers and downregulation of heat dissipation. Darkadapted ΦPSII was measured using the chlorophyll fluorometer. The plant was then exposed to white LED light with a photosynthetic photon flux density of approximately 550 µmol m−2 s−1 at the top of the canopy to induce upregulation of heat dissipation. After 15 min of light exposure, the LED fixture was removed from the imaging system and CFI images were collected at regular intervals over the following 15 min to determine whether the CFI images responded to the downregulation of heat dissipation. The plant was kept

Sensors 2021, 21, 2055

all photosystem II reaction centers and downregulation of heat dissipation. Dark-adapted ΦPSII was measured using the chlorophyll fluorometer. The plant was then exposed to white LED light with a photosynthetic photon flux density of approximately 550 µmol m−2 s−1 at the top of the canopy to induce upregulation of heat dissipation. After 15 min of light exposure, the LED fixture was removed from the imaging system and CFI image5sowf 1e7re collected at regular intervals over the following 15 min to determine whether the CFI images responded to the downregulation of heat dissipation. The plant was kept in the dark indtuhreindgartkhidsupreirnigodth, iesxpceepritofdo,retxhceebprtifeofrlitghhetberxiepfolsiughret reexqpuoisruerdefroerqtuhiereidmfaogrinthge. Fimolalogwinign.g Foelalcohwiimngageeacahcqimuiasgiteioanc,qtuhiesicthiolno,rothpehcyhllloflruooprhoymlleflteurowroams eutseerdwtoasmuesaesdutroemΦPeaSsIuI raenΦd PNSPIIQ, aninddNicPaQtiv, einodficthateivueporefgthuelautipornegouf lhaetiaotndoisfshipeaattiodnis.sNipPaQtiown.asNcPaQlcuwlaatsecdalbcausleadteodnbtahseedmoenasthuermedeafsluuorerdesflceunocreesdceunrcinegduthreinsgatthuerastaitnugraptiunlgsepaufltseeratfhteeritnhietiianlitdiaalrkdaarckclaicmclaimtioantio(Fnm()Famn)d anthdethmeemaseuarseudrefdlufloureosrceesncecencdeudruinrgintghethseastuatruartiantigngpupluselsseasfateftrerthtehe151-5m-mininuuteteeexxppoosusurereto tolilgighht t(F(Fmm’)’a) sas(F(mFm− F−m’F)/mF’m)’/.Fm’.
ThTeheavaevreargaegpeipxiexleilnitnetnesnitsyitaynadndstasntadnadrdarddedveiavtiiaotnioonf oaf5a05×0 ×5050pipxiexlealraeraeaclcolsoesteotothtehe fluflouroersecsecnecnecme emaesuasruemreemnetsnwtsaws aqsuaqnutaifinetidfiuedsinugsitnhge tFhieji sFoijfitwsoafrtewianrealilnCaFlIliCmFaIgiems aagnedswaend tewstedtefsotredcofrorrelcaotriorenlsatbioetnwsebeentwpeixeenl pinixteenl siinttye,nΦsPitSyI,IΦ, aPnSdIIN, aPnQd tNoPdQetteormdeinterwmhineethwerhCetFhIer imCaFgIeismcaoguelsdctorauclkd ctrhaacnkgcehsaingΦesPSinIIΦaPnSdIINaPnQd.NPQ.
3.3R. Reseuslutlsts 3.31..1C. ChlholroorpohpyhlyllFl lFulouroersecsecnecnecIemImagaignigng
AftAerftaepr paplipcalitcioatnioonf othfethaetraatzrianzein, ae,reaarseaws iwthitbhribgrhigthfltufoluroesrceescnecnecbeebceacmame eeveivdiednetnitnin thtehechchlolororopphhyyllllflfuluoorreesscceenncceeiimmaaggeess,, bbeeggiinnnniinngg aabboouuttttwwoohhoouursrsafatfetrerthtehaepapplpiclaictiaotnio.nIn. Inccrreeaasseeddflfluuooreresscceenncceeininitiitailallylywwasaesveivdiednetnatloanlogntghethmeimdviedivneoinf tohfetlheeavleeasvaensdalnardgelarragreeras aroefatshoefctahneocpaynfolupoyrflesucoerdesincetednisnetleynasseltyheasattrhaezianteraszpirneeadsptrheraodugthhroseucgohndseacroynvdeainrys (vFeiignusre (F1igAu).re 1A).

Figure 1. The effect of atrazine drench application on: (A) chlorophyll fluorescence and (B) visual Fiagpupreea1r.aTnhcee oefffCecatthoafraatnrtahzuinserodsreeunsc. hTiamppe lsicinactieohneornb:ic(iAd)ecahploprloicpahtiyolnl flisuoinrdesicceantecde ianndth(eBu) pvpiseural apperaigrahnthceanodf Ccoatrhnaerranotfheuascrhosiemuas.gTei.me since herbicide application is indicated in the upper righthand corner of each image.
Similar trends were seen in replications two and three (Supplementary Figures S2 and S3) as well as lettuce and new guinea impatiens (results not shown). In the RGB images, no evidence of any damage was visible for the first eight hours after herbicide application (Figure 1B). Damage became visible only around 36 to 48 h after herbicide application (Supplementary Figure S4).
3.2. Examination of Fluorescence from Reflectance Measurements To confirm that the application of atrazine caused changes in chlorophyll fluores-
cence intensity, changes in the normalized reflectance/fluorescence were analyzed. These changes over time depended on wavelength, with wavelengths from 640 nm to 750 nm

Sensors 2021, 21, 2055

6 of 17

exhibiting a relatively large increase in reflectance/fluorescence over the duration of the experiment (Figure 2). The changes in the measured reflectance/fluorescence at different wavelengths over the course of the study were quantified by calculating the standard deviation of the normalized reflectance/fluorescence at each wavelength across all time periods, with a higher standard deviation indicating larger changes in the normalized reflectance/fluorescence. The largest variation in the normalized reflectance/fluorescence Sensors 202o1,c2c1,uxrFrOeRdPEaERt RwEVaIvEWelengths above 650 nm, with peaks at 680 and 750 nm (Figure 3)7. oTf h18e standard deviation of the normalized reflectance/fluorescence was low and similar at wavelengths from 450 to 600 nm.

FiguFreig2u. Trhee2n.orTmhaeliznedorrmeflaecltiaznecde/frleufloreecscteannccee(n/oflrmuoalrizeesdcetontchee (mneoarsmureadlivzaeludetaot ttihmee 0m) oefaCsauthreardanvthaulsureoseauts taitme 0) diffeorefnCt watahvaerleanngtthhus sasraosfeuuncstiaotndoifftfheeretinmtewsinacveealteranzgintehaspapslicaatfiuonn. cVtaiolunesoinf tthheefitgiumreelesginencdesaintrdaiczaitne ethaepwpavliec-ation. lengVthaaltuwehsicihnththeereffilegcutarnecel/eflguoernedscsenicnedwicasamteeatshuerewd (anvme)l.eTnhge tthwoadt awtahseictshinthreed rinedfliceactteatnhecew/aflveuleonrgethscs eunsecde was
for the development of the fluorescence-based stress index.
measured (nm). The two data sets in red indicate the wavelengths used for the development of the
fluorescence-based stress index.

SSeennssoorrss 22002211,, 2211,, 2x0F55OR PEER REVIEW

7 of 187 of 18

FFigiguurere33. .TThheeeefffefecct tooffaatrtarazzinineeaapppplilciactaitoinonononthtehestsatnadnadradrddedveivaitaiotinonofotfhtehneonromrmaliazleizdeadvaevraegraege rreeflfelcetcatnacnec/efl/fuluoroersecsecnecnecoef oCfaCthaatrhaanrtahnutshruosseruossaesusa afusnactfiuonncotifotnheowf tahveelwenagvthel.eDnagttahu. sDeadttaoucsaelcdutlaotecalcuthlaetesttahnedsatradnddeavridatdioenviwataisontakweansetvaekreyn1e5vmeriny a1f5temr ihnerabfitceirdheearpbpicliicdaetiaopnpfolirca8thio(nnf=or328).h (n = 32).
3.3. Pixel Intensity, ΦPSII, and nARFS 3.3. Pixel Intensity, ΦPSII, and nARFS
Over the course of the experiment, the intensity of the 50 × 50 pixels area near the fluoroOmveteerr tahnedcroefluercsteanocfetmheeaesxupreermimenetnsti,tethinecrienatseends,itinydoicfatthivee5o0f i×nc5r0eapseixdeclshlaorreoaphnyelal r the flfuluooreroscmenetceer(aFnigdurreef4leAc)t.anTcheerme ewaassurveamrieanbtilistiyteinintchreeapsiexde,l iinndteincasittiyveamofoinngcrtehaesethdrecehlororpephlyiclal tfilounosr,ebsucet nthcee c(hFaignugreein4Ap)ix. eTlhienrteenwsiatys vfoalrloiawbeilditay sinimthilearppixaettleirnntefonrsiatlyl tahmreoenpglatnhtes.three Priexpeliicnatteinosnitsy, bwuatsthreelacthivaenlgyestinabpleixfeolritnhteenfisristtytwfoolltoowtherdeea hsiomuirlsa,rapftaetrtwerhnicfhoritagllrtahdrueaellpylants. inPcixreealsiendte. nLsikiteywwisaes, ΦrePlaStIiIvwelayssrtealabtlievfeolyr tshtaebfliersfot rtwthoetfiortsht rteweohtooufrosu,rahftoeurrws,haifctheriwt ghriacdhually tihnecrΦePaSseIIdg. rLaidkuewalilsyed, ΦecPreSaIIsewda(sFriegluarteiv4eBly).sAtasbllieghfotrdtihpeifnirnstAtRwFoStocfcouurrehdoudrusr,inafgtetrhwe hich inthiteiaΦl 2P1S5ImI ginraadftuear lalytradzeincereaapspeldica(Ftiiognuirne t4wBo).oAf thselitghhrteedrieppliincantiAonRsF. SThoicscwuarrsefdolldouwreindg the biyniatiastlr2o1n5gminincraefatseer iantrtahzeinneAaRpFpSlicdautrioinnginthtewroemofatihnedtehrroefetrheepl8i-chatpioenriso.dTh(Fisigwuares f4oCll)o, wed dbuyriangsttrhoensgaminectriemaseethinat tthhee pniAxeRlFinStednusritinygintchreearseemd aaninddΦerPSoIfI tdheecr8e-ahsepde.riod (Figure 4C), 3d.4u. rRinelgattihonesshaipms ebettiwmeeenthAavtertahgeepPiixxeell Iinntteennssitiyty, ΦinPcSrIeIa, saendd naAndRFΦSPSII decreased.
The primary goal of the study was to establish whether the pixel intensity acquired from the chlorophyll fluorescence images can be used to detect stress in the plant. Comparing the pixel intensity to ΦPSII verified that changes in pixel intensity are indeed indicative of physiological changes in the plants. There was a strong negative correlation between the pixel intensity and ΦPSII in all three replications (r < −0.90): as the pixel intensity decreased, the ΦPSII increased (Figure 5). This relationship was present, but quantitatively different among the three plants, indicative of the non-quantitative nature of our CFI approach.
To confirm that a change in pixel intensity was associated with a change in chlorophyll fluorescence, the relationship between the pixel intensity and nARFS was examined. A strong positive correlation between the pixel intensity and the nARFS was seen in all replications (r > 0.86; Figure 6), but once again this relationship differed among the three plants. There was a negative correlation between the nARFS and ΦPSII (r < −0.82, Figure 7), indicating that changes in nARFS were related to changes in the photosynthetic efficiency of the plant.

Sensors 2021, 21, 2055
Sensors 2021, 21, x FOR PEER REVIEW

9 of 18

8 of 17

FFiigguurree44. .ThTeheeffeecftfoefcattroafzianteraapzpinliecataiopnpolinc:a(tAio) ntheoanv:er(aAge) pthixeelainvteernasigtye opfiax5e0l pinixteelnbsyit5y0 of a 50 pixel by 5p0ixpeilxaerleaarneeaarnwehaerrwe thheeroeththeremoetahseurremmeeanstus rweemreetnatksenw, (eBr)etthaekqeuna,n(tBum) thyieelqduoafnPtSuIIm, (Cy)ieanldd of PSII, (C) and

tthhee nnoorrmmaalilzizeeddavaevreargaegreefrleecfltaenccteanocf ethoefrtehfleecrteaflnceectsapneccterusmpe(cntAruRmFS)(nofACRaFthSa)raonfthCuastrhoasreuans tohvuesr roseus over an

8a-nh8p-herpieordio.dT. hTheennAARRFFSS wwaassddeteetremrimneidnebdy abvyeraavgeinrgagthine gnotrhmeanliozerdmraeflliezcetdanrceeflinecthtaenccheloirno-the chlorophyll

phyll fluorescence spectrum (669.8–760.7 nm) and dividing that value by the value from the first

flruefolercetsacnecne cmeesapsuercetmruemnt (a6ft6e9r.h8e–r7b6ic0i.d7enampp)liacnatdiodni.viding that value by the value from the first reflectance

Sensors 2021, 21, x FOR PEER REVIEWmeasurement after herbicide application.

10 of 18

3.4. Relationships between Average Pixel Intensity, ΦPSII, and nARFS

The primary goal of the study was to establish whether the pixel intensity acquired from the chlorophyll fluorescence images can be used to detect stress in the plant. Comparing the pixel intensity to ΦPSII verified that changes in pixel intensity are indeed indicative of physiological changes in the plants. There was a strong negative correlation between the pixel intensity and ΦPSII in all three replications (r < −0.90): as the pixel intensity decreased, the ΦPSII increased (Figure 5). This relationship was present, but quantitatively different among the three plants, indicative of the non-quantitative nature of our CFI approach.

FFiigguurere5.5T.heTrheelatrieolnasthiiopnbsehtwipeebnetthwe eqeunanthume qyuiealdnotuf pmhoytoiesyldsteomf IpI hanodtothsye satveemragIeI painxedl the average pixel iinntteennsistiytyininchclohrlooprhoypllhfylullorfleuscoernecseciemnacgeesimofaCgaethsaorafnCthautshraorsaenusthtruesatreodsewuisthtraetaratzeidnew. ith atrazine.
To confirm that a change in pixel intensity was associated with a change in chlorophyll fluorescence, the relationship between the pixel intensity and nARFS was examined. A strong positive correlation between the pixel intensity and the nARFS was seen in all replications (r > 0.86; Figure 6), but once again this relationship differed among the three plants. There was a negative correlation between the nARFS and ΦPSII (r < −0.82, Figure

Sensors 2021, 21, 2055

To confirm that a change in pixel intensity was associated with a change in chlorophyll fluorescence, the relationship between the pixel intensity and nARFS was examined. A strong positive correlation between the pixel intensity and the nARFS was seen in all replications (r > 0.86; Figure 6), but once again this relationship differed among the three plants. There was a negative correlation between the nARFS and ΦPSII (r < −0.82, F9igouf r1e7 7), indicating that changes in nARFS were related to changes in the photosynthetic efficiency of the plant.

Figure 6. TThheerreellaattiioonnsshhiippbbeetwtweeeennnnoormrmaalilzizededavavereargaegererfelflecetcatnacneceofoffluflouroersecsecnecnecsepsepcetrcutrmumanadnd
Sensors 2021, 21, x FOR PEER REVIEWtwwhaaessattvaaekkreeanngeffrropomimxettlhhieenctcehehnnllososriirtotoyyppohohfyfyClClllaaflfttlhuhuaoaorrraraenensstcthcheuenunscscerreooismsiemeuuasagsgtetrerseesaianittneeadda5w5w00i×itt×hh55aa00ttraraaarrzezeiainannene.e.eTaaTrhrhtethehaeaevvsesperproaoatgtgoeoefpfptihtxihxeeeelrlreieniflnftelteecenctnas-sniticytye tmaneacesumreemaseunrtesm. ents.

11 o

FFigiguurree 77.. TThhee rreellaatitoionnshsihpipbebtweteweneethnethnoernmoarlmizeadliazveedraagveerreaflgeectraenfcleecotfaflnucoeroesfcfelnucoersepscecetnrucemspectrum (n(nAARRFFSS)) aannddththeequqaunatunmtuymielydieolfdphooftposhyosttoemsyIsItoefmCaItIhoarfaCntahtuhsarroasneuths utrseartoesdeuwsitthreaatrteazdinwe.ith atrazine.
3.5. Development and Support for a Fluorescence-Based Stress Index
3.5. DTehveelFoBpSmIe(nEtqaunadtioSnu(p1p)o) rwt afosrdaevFelluoopreedscbenasceed-BoansetdheStsrtaesnsdIanrddedxeviation of the reflectaTnhceemFBeaSsIu(reEmqeunattsiaotnd(if1fe))rewntaws adveevleenlgotphesd(Fbigausreed3)o. Tnhtehlearsgteasnt sdtaanrddadrdevdieavtiiaotinonosf the ref tainnmceeamsueraesdurreeflmecetanntcse/aflt udoirfefescreennctewocacvuerrleednagtth69s8(.3Fingmuraend3)7.4T1.h2enmla,rwgehsiltesthtaenlodwaredst deviati
standard deviation occurred at 548.9 nm. The reflectance measurements at 741.2 nm were
inchmoseeansfuorreudserienftlheectFaBnScIea/sfltuheoyrewsecreenbceettoercccourrrreeladteadtw6i9th8.Φ3 PnSmII tahnadn t7h4e1m.2eansmur,ewmheniltes the low standard deviation occurred at 548.9 nm. The reflectance measurements at 741.2 nm w chosen for use in the FBSI as they were better correlated with ΦPSII than the measu ments at 698.2 nm, while the reflectance at 548.9 nm was used for normalization. The F was defined using these wavelengths, where R is the reflectance at a specific waveleng

Sensors 2021, 21, 2055

3.5. Development and Support for a Fluorescence-Based Stress Index

The FBSI (Equation (1)) was developed based on the standard deviation of the reflectance measurements at different wavelengths (Figure 3). The largest standard deviations in measured reflectance/fluorescence occurred at 698.3 nm and 741.2 nm, while10tohfe17lowest standard deviation occurred at 548.9 nm. The reflectance measurements at 741.2 nm were chosen for use in the FBSI as they were better correlated with ΦPSII than the measuremenatts6a98t.629n8m.2, nwmhi,lewthhieleretflheectraenfcleecatta5n4c8e.9ant m54w8.a9snumsewd faosrunsoerdmfaolirzantoiornm. aTlhizeaFtBioSnI .wTahse FBSI wasddefienfiendeudsuinsginthgetsheewsaevwelaenvgetlhesn,gwthhse,rewRhiesrteheRriesfltehctearnecfeleact taasnpceeciafitcawsapveecleifnigctwh:avelength:

FBSFIB=S(IR=74(1R.2 7−41.2R–54R8.954)8/.(9R) 7/41(.R2 7+41R.2 5+48R.95)48.9)

(1) (1)

The FBSI was calculated for each timepoint and there were strong, negative correla-
The FBSI was calculated for each timepoint and there were strong, negative correla-
tiontsiobnestbweteweneetnhteheFFBBSSIIaanndd ΦΦPPSSIIIIffoorralalltlhtrhereereeprleicpaltiicoantsio(nr s< (−r0<.8−, 0F.i8g,uFreig8u),rbeu8t )o,nbcuet once agaaingatihnetsheesceocrorerrlealtaitoionnsswweerree qquuaannttiittaatitviveleylydidffiefrfeenrteanmt aomngotnhge tthhreeethprlaenetps.lants.

FiguFriegu8r.eT8h. eThreelaretliaotniosnhsihpipbebtewtweeeennththeefflluuoorreesscceennccee-b-baasesdedstsretrsessisndinedx eaxndanthdetqhueaqnutuamntyuimeldyoiefld of phoptohsoytossteymstemII oIIfoCf aCtahtahraarnanththuussrroosseeuussttrreeaatetdedwwithitahtraatzrianzei.ne.
3.6. Pixel Intensity, Heat Dissipation, and Quantum Yield Recovery in the Dark
The petunia plant fluoresced notably more brightly when dark-adapted than following a 15-minute exposure to a photosynthetic photon flux density of 550 µmol m−2 s−1 (Figure 9). The fluorescence intensity increased gradually following the light exposure, from a pixel intensity of 54.6 ± 3.8 at 30 s after the end of the light exposure to 59.6 ± 3.6 after 15 min. This increase in fluorescence was not clearly visible in the CFI images, but easily quantified. Pixel intensity was negatively correlated with NPQ and positively correlated with ΦPSII. Both correlations were highly significant, indicating that changes in pixel intensity in CFI images can be used to qualitatively follow downregulation of NPQ along with the associated increase in ΦPSII.
Pixel IntensityΦpsiiPlantPlantsNarfs