The Use of High-Throughput Phenotyping for Assessment of Heat Stress-Induced Changes in Arabidopsis

The worldwide rise in heatwave frequency poses a threat to plant survival and productivity. Determining the new marker phenotypes that show reproducible response to heat stress and contribute to heat stress tolerance is becoming a priority. In this study, we describe a protocol focusing on the daily changes in plant morphology and photosynthetic performance after exposure to heat stress using an automated noninvasive phenotyping system. Heat stress exposure resulted in an acute reduction of the quantum yield of photosystem II and increased leaf angle. In longer term, the exposure to heat also affected plant growth and morphology. By tracking the recovery period of the WT and mutants impaired in thermotolerance (hsp101), we observed that the difference in maximum quantum yield, quenching, rosette size, and morphology. By examining the correlation across the traits throughout time, we observed that early changes in photochemical quenching corresponded with the rosette size at later stages, which suggests the contribution of quenching to overall heat tolerance. We also determined that 6 h of heat stress provides the most informative insight in plant's responses to heat, as it shows a clear separation between treated and nontreated plants as well as the WT and hsp101. Our work streamlines future discoveries by providing an experimental protocol, data analysis pipeline, and new phenotypes that could be used as targets in thermotolerance screenings.

The worldwide rise in heatwave frequency poses a threat to plant survival and 23 productivity. Determining the new marker phenotypes that show reproducible 24 response to heat stress and contribute to heat stress tolerance is becoming a priority. 25 In this study, we describe a protocol focusing on the daily changes in plant 26 morphology and photosynthetic performance after exposure to heat stress using an 27 automated non-invasive phenotyping system. Heat stress exposure resulted in an 28 acute reduction of quantum yield of photosystem II and increased leaf angle. In the 29 longer term, exposure to heat also affected plant growth and morphology. By 30 tracking the recovery period of WT and mutants impaired in thermotolerance 31 (hsp101), we observed that the difference in maximum quantum yield, quenching, 32 rosette size, and morphology. By examining the correlation across the traits 33 throughout time, we observed that early changes in photochemical quenching 34 corresponded with the rosette size at later stages, which suggests the contribution 35 of quenching to overall heat tolerance. We also determined that 6h of heat stress 36 provides the most informative insight in plant responses to heat, as it shows a clear 37 separation between treated and non-treated plants as well as WT and hsp101. Our 38 work streamlines future discoveries by providing an experimental protocol, data 39 analysis pipeline and new phenotypes that could be used as targets in 40 thermotolerance screenings. 41 42 43 44 1 1. Introduction 2 3 Globally, the last decade was the warmest since the 19th century, and resulted in record-4 breaking heat waves in many parts of the world (Coumou et al., 2013;Schiermeier, 2019). 5 Heat stress leads to a reduction of plant performance and productivity at all developmental 6 stages, making the heatwaves a serious threat to agriculture. However, the majority of the 7 efforts in heat stress research focus either on early seedling development, scoring survival 8 or hypocotyl elongation (Larkindale et al., 2005;Silva-Correia et al., 2014;McLoughlin et 9 al., 2016) or reproductive stages (Zhou et al., 2017;Fan et al., 2018), where the pollen 10 viability is reduced by high temperatures. The handful of studies focusing on the heat stress 11 responses at the vegetative development stage (Rodríguez et al., 2015;Xu et al., 2017;12 Cheabu et al., 2018) show that heat tolerance at vegetative stage contributes to resilience 13 at the reproductive stage. Therefore, understanding the changes caused by heat stress and 14 breeding for heat tolerance at all developmental stages is essential to ensure future 15 sustainable food supply. 16 17 Arabidopsis thaliana has been widely used in screenings for thermotolerance, 18 predominately focusing on seedling viability (Li et al., 2007;Gao et al., 2008;Suzuki et 19 al., 2008), hypocotyl elongation (Hong and Vierling, 2000;Charng et al., 2007), or seed 20 germination (Silva-Correia et al., 2014) on agar plates. As heat tolerance relies on multiple 21 processes, quantification of simple traits, determined by the ease of phenotyping rather 22 than physiological importance, does not provide the best tools capturing the complexity of 23 the responses, e.g. plant cooling capacity, growth recovery, and maintenance of 24 photosynthesis, which all contribute to the diversity of thermotolerance mechanisms. 25 Continuous monitoring of plant growth after heat exposure via non-destructive methods, 26 such as RGB, thermal imaging and chlorophyll fluorescence, provide insight into 27 physiological responses corresponding to photosynthetic efficiency and plant cooling 28 abilities which cannot be scored by eye. The automated and environmentally controlled 29 system enables time-efficient screening of large populations in a single experiment. 30 Phenotypic traits, such as plant size, temperature, and photosynthetic efficiency have been 31 successfully applied to evaluate plant performance under drought (Jansen et al., 2009;Chen 32 et al., 2014), salinity (Awlia et al., 2016), and chilling (Jansen et al., 2009) stress, but to 33 our knowledge, no such study has been conducted on study heat stress response. 34 35 HSP101, a molecular chaperone involved in protein disaggregation was one of the earliest 36 genes identified to have a crucial role in thermotolerance in Arabidopsis, with no 37 detrimental effects on normal growth or development in the absence of stress (Queitsch et 38 al., 2000;Hong and Vierling, 2001). Homologs of HSP101 were identified and 39 characterized for their role in heat stress response in maize, soybean, wheat, tobacco and 40 pea, kidney bean (Keeler et al., 2000;Katiyar-Agarwal et al., 2001). As such, hsp101 41 mutant showed a severe reduction in heat tolerance compared to wild-type in terms of 42 survival, however the broader knowledge about the processes compromised in this mutant 43 during the heat exposure are unknown. 44 45 In this study, we investigate the feasibility of applying RGB, kinetic chlorophyll 46 fluorescence and infrared imaging for evaluating heat stress response in Arabidopsis. We 47 developed a physiologically relevant heat-imposition protocol for the vegetative stage of 48 3 Arabidopsis plants based on the significant changes observed for multiple traits. 1 Additionally, by studying the heat-induced changes in WT and hsp101 mutant, we were 2 able to identify additional traits that might indicate compromised heat stress tolerance. By 3 applying machine learning, we identified that maintenance of photochemical quenching 4 immediately after stress application could be potentially used as an indicator for heat stress 5 tolerance, as it corresponded with the increase in plant size at the later time points. This 6 work provides a primer for future studies using high-throughput phenotyping platforms, 7 uncovering novel components of heat stress tolerance. 8 9 10 2. Materials and Methods 11 12 2.1 Plant materials and growth conditions 13 Seeds of Arabidopsis wildtype Col-0 (CS60000) and hsp101 (AT1G74310; hot 1-3, NASC 14 ID: N16284) were stratified for three days at 4°C in the dark and germinated in controlled 15 environment in PSI growth room (Photon Systems Instruments, Czech Republic). The 16 environmental setting of PSI growth room) was set at 22°C (sensor sensitivity range: ± 17 0.1°C), with a relative humidity of 60% (sensor sensitivity range: ± 1%) and 400 ppm 18 (sensor sensitivity range: ±100 ppm) of CO2. At four-leaf stage (day 14 after sowing), 19 healthy seedlings with similar size were transferred into PSI standard pots (6 cm x 6 cm x 20 9.5 cm) filled with 100 g (± 1.0 g) of the growing mix (SunGro Horticulture Metro-Mix 21 360, MA, USA), placed into PSI trays (5 x 4 pots per tray) and registered into the 22 PlantScreen TM system. All pots were automatically weighed and watered every day to reach 23 and maintain the weight of 130 g. Plants were grown under cool-white LED panel with a 24 16 h/8 h light/dark cycle, the light intensity received at plant rosette level is ~150 μmol m -25 2 s -1 . All plants were kept in the PSI growth room during the experiment except during the 26 heat treatment. 27 28

Heat stress treatment and phenotyping experimental design 29
At day 22 after sowing, we subjected plants to 3 h, 6 h or 9 h of heat stress and control 30 treatments (Figure 1). For each treatment group (3 h, 6 h, 9 h or control), there are two 31 trays each containing 10 wildtype Col-0 and 10 hsp101 plants next to each other in an 32 evenly distributed design. Heat stress was applied by placing the 9 h, 6 h and 3 h treatment 33 trays into a pre-heated Percival growth chamber (Model CU36-L5, Percival Scientific, IA, 34 USA) with white lights on (45°C, ~120 μmol m -2 s -1 ) from 9 am to 6 pm, from 12 pm to 6 35 pm, and 3 pm to 6 pm respectively. Two trays used as "control treatment" with the same 36 composition of genotypes remained at the PSI growth room. After heat stress application, 37 six trays were transferred back to the PSI growth room. The transfer of the trays between 38 the phenotyping facility and the heated growth chamber lasted between 2-3 minutes. Plants 39 were imaged using chlorophyll fluorescence, RGB and infrared cameras daily at 7 pm 40 starting from the day before the heat stress application (DAS -1) until one week after 41 (Figure 1). Using image-based analysis, we acquired a variety of traits reflecting plant 42 growth, photosynthetic efficiency, rosette morphology, and temperature in Col-0 and 43 hsp101 plants and explored these phenotypes for phenotypic plasticity in response to heat 44 stress and their feasibility for large thermotolerance screening. In total, we screened 160 45 individual plants, with 20 biological replicates per genotype per treatment, to develop an 46 4 understanding of changes induced by treatment, genotype and interaction between them 1 for various traits. 2 3 2.3 Imaging-based phenotypic measurements 4 Plant imaging was initiated one day before the heat application (DAS -1) to provide a 5 baseline for the analysis and was performed daily at 7 pm until 7 days after the heat 6 application (DAS 7). Each imaging round consisted of an initial 15 min dark-adaptation 7 period inside the acclimation channel, followed by chlorophyll fluorescence, red green blue 8 (RGB) colored and infrared (IR) imaging. For each imaging round, the phenotyping time 9 for all trays was about 80 mins. Lighting conditions, plant positioning, and camera settings 10 were fixed throughout the experiment. 11 Chlorophyll fluorescence imaging unit in PlantScreen TM Systems constructed by Photon 12 System Instruments (PSI, The Czech Republic) measures the re-emitted light 13 approximating the photosynthetic performance of plants' photosystem II. The light curve 14 protocol from PSI was applied to provide detailed information on fluorescence kinetics 15 during the heat stress recovery (Henley, 1993;Rascher et al., 2000), a detailed protocol 16 can be found in Figure S1. After 15 minutes of dark adaptation, the initial flash of light 17 was applied to measure the minimum fluorescence (Fo), followed by a saturation pulse to 18 determine the maximum fluorescence (Fm) in the dark-adapted state. 3.1 Extended exposure to heat stress results in a proportional decrease of the rosette 9 size and photosynthetic efficiency 10 11 To assess whether high-throughput phenotyping can capture significant alterations in plant 12 physiology caused by exposure to heat stress, we exposed three weeks old Arabidopsis 13 plants to 3 h, 6 h or 9 hours of acute heat stress (45°C), and evaluated the plant size, 14 morphology, temperature, and chlorophyll fluorescence for the subsequent 7 days after 15 stress (DAS) imposition (Figure 1). All three heat stress treatments resulted in a significant 16 reduction of the rosette size ( Figure 2B), and these differences were substantial already 1 17 h after heat stress treatment. In general, the extended exposure to high temperature 18 increased the effect observed on the rosette size ( Figure 2B). The applied treatments were 19 sub-lethal to the plants, as only four plants died after 9 h of exposure to heat stress, while 20 other 36 plants that underwent this treatment were able to recover from the stress and 21 produced new green leaves. These results suggest that soil-grown 3-week-old Arabidopsis 22 plants are resilient to acute heat stress and that 6h of heat stress treatment is best at 23 differentiating between WT and heat-sensitive mutant hsp101( Figure 2C). The phenotyping protocol. Each tray underwent an initial 15 min dark-adaptation period inside the 6 adaptation chamber, followed by chlorophyll fluorescence, red green blue (RGB), and thermal 7 imaging, with automatic weighing and watering before returning to the growth chamber. (C) The 8 heat stress imposition protocol. 22 days after sowing, two trays of plants were kept in the growth 9 chamber as control and the other six trays were moved into the pre-heated 45°C Percival chamber 10 at 9 am, 12 pm and 3 pm for the 9 h, 6 h, and 3 h heat treatment respectively. All treated six trays 11 were returned to the growth chamber at 6 pm and imaged daily at 7 pm starting from the day before 12 the heat stress application (DAS -1) until one week after. (D) The overview of the chlorophyll 13 fluorescence protocol executed using the dark-adapted plants. The minimal (Fo) and maximal (Fm) 14 fluorescence are measured directly after dark adaptation, followed by gradual exposure to 15 increasing light intensities of 95, 210, 320, 440, 555 and 670 µ mol m -2 s -1 , corresponding to Lss 1, 16 2, 3, 4, 5, and 6 respectively, where the minimal (Fo') and steady-state fluorescence are determined.

17
At each light intensity plants are exposed to a saturating light flash, which allows measuring the 18 maximum fluorescence at light-adapted state for given intensity (Fm').

20
To further evaluate the changes in relationship to overall heat tolerance, we compared Col-21 0 (WT) and hsp101 mutant (Figure 2A, C). After heat stress imposition, the hsp101 plants 22 developed smaller rosettes sizes than WT (Figure 2C), while no significant difference in 23 rosette size was observed between WT and hsp101 without heat stress imposition ( Figure  24 S1A). The significant differences in rosette size were observed between Col-0 and hsp101 25 two days after the 3 h heat treatment and three days after the 6 h treatment. No difference 26 7 between Col-0 and hsp101 was observed after the 9 h treatment, where the rosette size was 1 reduced in a similar degree ( Figure 2C) .  2  3  While plant size broadly reflects overall plant performance, high temperature can also  4 influence plant morphology, such as the petiole elongation and increased leaf angle (Koini 5 et al., 2009;Crawford et al., 2012). We examined the effect of heat stress on rosette 6 morphology and observed that heat stress treatment had a pronounced effect on rosette 7 perimeter, compactness, and slenderness of leaves already one hour after heat stress 8 application (Figure S2 A). In addition, these three parameters showed significant 9 differences between heat stress treated Col-0 and hsp101 lines (Figure S2 B). Another 10 morphological change that we observed one day after the heat application was the increase 11 in leaf angle, which occurred only in Col-0 plants but not in hsp101 (Figure S2 D). The 12 change in leaf angle was reflected by the transient increase of rotational mass symmetry in 13 Col-0 at DAS 1, but returned back to levels observed in non-stressed plants 2 days after 14 stress application (Figure S2 B).
The decrease in plant photosynthetic efficiency is generally believed to precede 17 developmental changes. Based on the chlorophyll fluorescence we indeed observed a 18 decline in maximum quantum yield (QY max), derived from the measurements of dark-19 adapted minimum (Fo, Figure S3) and maximum fluorescence (Fm, Figure S4) as (Fm -Fo) 20 /Fm. QY max indicates the efficiency of PSII photochemistry in the dark-adapted state and 21 reveals the efficiency of electron transport inside PSII. The immediate decline in QY max 22 occurred on the stress day (DAS 0) for both WT and hsp101 (Figure 2 D-E). The WT was 23 able to recover the QY max 1 DAS to levels observed in non-stressed plants, whereas QY 24 max of hsp101 plants recovered only at 2 DAS (Figure 2 E). We also noted the severe 25 reduction of the QY max in hsp101 at the edges of the rosette (Figure 2 A), which preceded 26 the tissue senescence. The earlier reduction in QY max in these areas suggests that change 27 in chlorophyll fluorescence can be used as early indicators of premature senescence. Exposure to high-temperature was earlier reported to result in an instant increase in 5 minimal (F0) and the decrease in maximum fluorescence (Fm) in various plant species 6 (Schreiber and Armond, 1978;Yamane et al., 1997). The light-curve protocol used to 7 examine the chlorophyll fluorescence, allowed us to study heat-induced changes in light-8 adapted state at various light intensities (Figure 1 D). Heat stress exposure resulted in 9 lower Fm measured at 0 DAS ( Figure S4), while the significant decrease for Fo was 10 observed at 1 DAS in WT plants for all three heat stress regimes (Figure S3 A). By 11 examining all the directly measured traits at 0 DAS (Figure 3 A), we noted a heat-stress 12 induced decrease in light-adapted Fm at all studied light intensities. This reduction in 13 maximum fluorescence was more pronounced in hsp101 plants for the lowest and highest 14 light intensities studied (Figure 3 B). The heat stress also resulted in reduced minimal and 15 steady-state fluorescence (Ft) across studied light intensities at 1 DAS (  As plant's cooling capacity is expected to be affected by exposure to high-temperature, we 17 examined differences in leaf temperature recorded from an infrared camera between heat-18 stressed and non-stressed plants ( Figure S5). Heat exposure increased leaf temperature 19 across three different heat stress treatments (Figure S5 A), no significant differences in 20 leaf temperature were observed between WT and hsp101 plants (Figure S5 B).

22
3.2 Classification model and trait selection to differentiate heat-sensitive and 23 tolerant lines 24 25 As the high-throughput phenotyping dataset allows us to examine more than 80 phenotypes, 26 it makes it difficult to rank individual phenotypes to select the best traits which allow clear 27 differentiation of heat-sensitive genotypes. Therefore, we implemented the machine 28 learning to simultaneously explore all the quantitative phenotypes collected within our 29 experiment to differentiate between WT and hsp101 mutant. We applied logistic regression 30 with lasso regularization to select the most useful traits for classification. To identify the 31 **** **** **** **** **** **** ** **** **** **** **** * 6 h heat stress − 0 DAS B **** *** **** **** **** **** **** **** **** **** **** **** **** **** **** *** **** **** **** heat stress regime allowing us to differentiate between WT and hsp101, we compared the 1 classification performance of phenotypic data from different treatment groups including 2 rosette size, leaf temperature and a subset of top morphological traits (perimeter, the 3 slenderness of leaves, compactness, isotropy, and rotational mass symmetry) and 4 chlorophyll fluorescence parameters measured in dark-adapted state (Fo, Fm, and QY max). 5 The accuracies of model prediction (Table 1) were modest, with the highest accuracy of 6 68.2% for the 3 h heat treatment. The predictions calculated for 9 h heat stress treatment 7 was lower than for the non-stressed plants, implying that 9 h heat stress treatment was too 8 severe and not suited to differentiate phenotypic differences between the genotypes. When 9 we included all phenotypic traits in the model, including the chlorophyll fluorescence 10 measured in light adapted state, we observed improved classification accuracies for all 11 groups ( Table 1). In line with our earlier analysis (Figure 2) 1  susceptibility  2  3 As photochemical quenching (Fq) was a unique trait that was associated with differentiation 4 between the WT and heat-sensitive hsp101 mutant, we examined the heat-induced changes 5

Decline in photochemical quenching as an early indicator of heat stress
in Fq throughout the duration of the experiment (Figure 4). The heat exposure resulted in 6 an immediate reduction of Fq across all heat stress regimes, and recovery to the Fq levels 7 observed for the non-treated plants within 1, 2 or 3 DAS for WT plants exposed to heat 8 stress for 3, 6 and 9 h respectively (Figure 4 A). The heat-sensitive hsp101 mutant 9 showed an even more severe decrease in Fq immediately after heat stress exposure in plants 10 exposed to heat stress for 3 and 6 hours (Figure 4 B). While we also observed a reduction 11 in the non-photochemical quenching measured at the same light intensity (Lss5) for the 12 WT plants exposed to heat stress (  In order to examine whether the heat stress-induced reduction in Fq was corresponding to 1 the plant's performance at the later stage of the experiment, we examined the correlations 2 for individual plants between Fq and rosette area measured at individual DAS ( Figure 5). 3 For plants that did not undergo the heat stress treatment, we observed a correlation between 4 Fq and area across all time points (Figure 5 A), however, the correlation within the traits 5 (e.g. area at different DAS) was higher than between the traits. Interestingly, for the plants 6 exposed to heat stress for 3 and 6 h, the Fq measured at 0 and 1 DAS shows strong 7 correlations with the rosette area measured throughout the experiment (Figure 5 B-8 C). This suggests that the maintenance of Fq could be used as an indicator of plant 9 performance. We also examined the correlations throughout the time between other traits 10 that ranked high in logistic regression classification ( Table 1) and the rosette area. The 11 heat stress-induced changes in isotropy were negatively correlated with rosette area at 12 earlier time points (DAS 2-4, Figure S7). The rosette compactness at 1 DAS also exhibited 13 a small negative correlation with the rosette area at later stages (5-7 DAS) ( Figure S8). 14 The slenderness of leaves (SOL) scored at early timepoints after stress was not correlated 15 with the plant area at 3 or 6 h of heat stress, and at 9 h the rosette area scored at earlier time 16 points was negatively correlated with SOL, suggesting that observed changes in SOL are 17 rather a consequence of reduced rosette area, rather than the cause ( Figure S9). 18 19 14 1 Figure 5. Early changes in photochemical quenching correspond to a larger rosette area. The 2 correlation matrix between photochemical quenching (Fq) and rosette area scored at various days 3 after stress (DAS) application for the plants (A) not-exposed to heat stress (B) exposed to 3 h, (C) As correlations between rosette area and compactness or isotropy were only weakly 12 significant (0.01 < p-value < 0.05), and the trends were only observed in plants exposed to 13 heat stress for 3 h, the correlation between early changes in Fq and rosette area size suggests 14 that maintenance of Fq might be causal to plant performance at a later stage. To examine 15 these correlations in more details, the Fq at 0 and 1 DAS was plotted for individual heat 16 stress regime and the correlation between the rosette area in individual genotypes was 17 examined (Figure 6). The correlation between Fq and area, when both traits are scored at 18 0 DAS, is weak and non-significant in most cases (Figure 6 A). However, when we 19  examined the correlation between Fq scored at 0 DAS with rosette area at later timepoints, 1 the correlation coefficients increased for the heat-stress treated samples (Figure 6 B, 2 Figure S10). Interestingly, across the increasing heat stress exposure, the correlations 3 between Fq at DAS 0 and rosette area at later DAS increased more decidedly for the heat 4 susceptible hsp101, suggesting that maintenance of photochemical quenching had even 5 higher importance for this heat susceptible genotype. Similar correlations were for Fq 6 measured at 1 DAS, where the correlations between Fq and rosette area measured both at 7 1 DAS were relatively weak (Figure 6 C), and increase when the rosette area is measured 8 at later time points (Figure 6 D, Figure S11). The Fq measured at 7 DAS also shows 9 significant correlations with the rosette area measured at the same time (Figure 6 E), 10 however the correlation coefficients are lower and the p-values are higher than for the 11 correlations between early changes in Fq and later size of the rosette area. These results are 12 strongly suggestive that photochemical quenching might be an important component of 13 heat stress tolerance, and that maintenance of Fq is particularly important in the heat 14 susceptible lines. 15 16 16 1 Figure 6. Heat-stress induced reduction in photochemical quenching indicates heat 2 susceptibility. The correlation between photochemical quenching (Fq) and rosette area was 3 examined for plants not-exposed to heat stress, and plants exposed to 3 h, 6 h or 9 h of heat stress 4 (45 °C). The correlation was examined between Fq scored at 0 days after stress (DAS) imposition Increasing temperature is one of the most important environmental factors affecting the 4 agricultural productivity worldwide. Improving our understanding of the mechanisms 5 underlying plant heat stress responses will facilitate the development of technologies and 6 breeding strategies for improving plant thermotolerance. In this manuscript, we show an 7 example of how high-throughput phenotyping can be used to screen for heat tolerance 8 related traits, providing more insight into the physiological processes contributing to 9 thermotolerance. 10 11 By studying the heat-induced changes in plant size, morphology, temperature and 12 chlorophyll fluorescence we identified a set of phenotypes like leaf temperature, maximum 13 quantum yield, and slenderness of leaves that show immediate response to heat stress. By 14 evaluating three different heat stress regimes (3, 6 and 9 h at 45°C , Figure 1), we identified 15 which trends were observed across all treatments. By including a heat-sensitive hsp101 16 mutant, we were able to distinguish which phenotypes were informative in distinguishing 17 between WT and heat sensitive lines ( Table 1). We observed that hsp101 plants showed a 18 more severe decrease in plant growth and quantum yield compared to the WT plants, and 19 that the difference between genotypes were most evident across the plants exposed to heat 20 stress for 6 h (Figure 2). The rapid change in quantum yield seems to be specific to heat 21 stress, as Arabidopsis plants exposed to salt stress did not show reduced photosynthetic 22 yield with similar length of the experiment (Awlia et al., 2016), while drought decreased 23 quantum yield only at the later stage of stress exposure (Jansen et al., 2009). While the 24 changes in quantum yield and photochemical quenching was observed immediately after 25 the heat stress, the heat stress also affected rosette morphology at later time points, 26 including the reduction in the slenderness of leaves, compactness and increased rosette 27 isotropy ( Figure S2). Such changes in the rosette morphology were less apparent under 28 salt stress (Awlia et al., 2016). These differences between heat, drought and salt reflect the 29 nature of abiotic stress, with heat stress exposure being the most acute, while the severity 30 of salt and drought stress increases gradually over longer periods of time. The observed 31 differences highlight the importance of selecting the physiologically relevant levels of 32 stress, considering the differences in the nature of individual abiotic stress that the plants 33 are exposed to the environment. Summarizing, results presented in this study showed that 34 plant physiological responses to high temperature are complex and temporal in their nature, 35 with short term changes being captured best with chlorophyll fluorescence (Fm, F0, and QY 36 max, Fq) and leaf temperature, while heat-induced changes in rosette morphology (rosette 37 area, perimeter, compactness, and slenderness of leaves) are observed at extended time 38 after the application of heat stress. While the majority of the heat stress studies focuses on 39 survival, we think that using the combination of these parameters and screening them at a 40 continuous time-window will provide a better understanding of processes underlying heat 41 tolerance. 42

43
In this study, we observed that while heat stress exposure increased the leaf angle in Col-44 0, this response was absent in hsp101 mutant lines ( Figure S3). While the response is 45 obvious to the human eye, it proved difficult to detect it from the available morphology 46 parameters, which use only top view images. The increased leaf angle was earlier suggested 47 to have an adaptive advantage under high temperatures, increasing the cooling capacity of 1 the plant (Crawford et al., 2012;Bridge et al., 2013). As we did not observe any significant 2 differences in leaf temperature between WT and hsp101 mutants (Figure S5), the cooling 3 advantage of this response is yet to be demonstrated. Nevertheless, the hsp101 showed a 4 greater reduction in maximum quantum yield compared to WT (Figure 2), and a lower 5 decrease in photochemical and non-photochemical quenching (Figure 4). However, as the 6 change in leaf angle was only observed 1 DAS, it is unlikely that this response to causal to 7 the decreased photosynthetic efficiency. How an increase in leaf angle and photosynthetic 8 quenching mechanisms are orchestrated by HSP101 is beyond the scope of this study. If 9 the leaf angle is to be used for future assays, it is our suggestion to include the 3D 10 measurements of the rosettes using the 3D scanner technology and/or side-view image. 11 12 While high throughput phenotyping provides more information on plant performance using 13 the non-destructive measurements, the number of the collected direct and derived 14 measurements ( which would enable to differentiate between WT and heat sensitive hsp101 mutant (Table  20 1). The results supported our earlier analysis that 6 h heat treatment is most informative for 21 differentiate between WT and heat sensitive genotype, and highlighted the significance of 22 morphology traits, direct chlorophyll fluorescence measurements as well as photochemical 23 quenching (Table 1). These results lead us to investigate changes in photochemical 24 quenching in more detail (Figure 4). We observed that maintenance of photochemical 25 quenching was positively correlated with larger rosettes at later timepoints ( Figure 5) and 26 that this correlation was predominantly found in the plants exposed to heat stress (Figre 6).

27
The contributions of non-photochemical quenching to heat stress tolerance (Havaux et al.,  28 1988), as well as other abiotic stress (Flagella et al., 1995) are widely described in earlier 29 literature. The photochemical quenching is related to the redox state of the first electron 30 acceptor of Photosystem II (Schreiber et al., 1986), but how this process is affected by 31 stress and what is its contribution to overall environmental stress tolerance remains 32 unknown. While we want to stress that the correlation does not prove causation, the 33 observed correlation between photochemical quenching and thermotolerance will be an 34 important cornerstone in future research of heat stress responses. 35 36 Using kinetic chlorophyll fluorescence assays, such as light curve protocol (Henley, 1993;37 Rascher et al., 2000) used in this study, provided new insights in dynamic changes to 38 photosynthetic efficiency and heat stress-induced photochemical quenching. We think that 39 the new phenotypic traits, presented in this manuscript will provide better insight and 40 identify novel players contributing to overall plant performance and heat stress tolerance. 41 As processes contributing to overall thermotolerance are complex (Kotak et al., 2007)  for assistance with the phenotyping facility.   Figure S1. Col-0 and hsp101 plants are indistinguishable when not exposed to heat         Days After Stress QY max (Fv/Fm) 9 h heat stress A **** **** **** **** **** **** ** **** **** **** **** Chlorophyll fluorescence (a.u.) Genotype a a