Simulation of Photosynthetic Quantum Efficiency and Energy Distribution Analysis Reveals Differential Drought Response Strategies in Two (Drought-Resistant and -Susceptible) Sugarcane Cultivars

Selections of drought-tolerant cultivars and drought-stress diagnosis are important for sugarcane production under seasonal drought, which becomes a crucial factor causing sugarcane yield reduction. The main objective of this study was to investigate the differential drought-response strategies of drought-resistant (‘ROC22’) and -susceptible (‘ROC16’) sugarcane cultivars via photosynthetic quantum efficiency (Φ) simulation and analyze photosystem energy distribution. Five experiments were conducted to measure chlorophyll fluorescence parameters under different photothermal and natural drought conditions. The response model of Φ to photosynthetically active radiation (PAR), temperature (T), and the relative water content of the substrate (rSWC) was established for both cultivars. The results showed that the decreasing rate of Φ was higher at lower temperatures than at higher temperatures, with increasing PAR under well-watered conditions. The drought-stress indexes (εD) of both cultivars increased after rSWC decreased to the critical values of 40% and 29% for ‘ROC22’ and ‘ROC16’, respectively, indicating that the photosystem of ‘ROC22’ reacted more quickly than that of ‘ROC16’ to water deficit. An earlier response and higher capability of nonphotochemical quenching (NPQ) accompanied the slower and slighter increments of the yield for other energy losses (ΦNO) for ‘ROC22’ (at day5, with a rSWC of 40%) compared with ‘ROC16’ (at day3, with a rSWC of 56%), indicating that a rapid decrease in water consumption and an increase in energy dissipation involved in delaying the photosystem injury could contribute to drought tolerance for sugarcane. In addition, the rSWC of ‘ROC16’ was lower than that of ‘ROC22’ throughout the drought treatment, suggesting that high water consumption might be adverse to drought tolerance of sugarcane. This model could be applied for drought-tolerance assessment or drought-stress diagnosis for sugarcane cultivars.


Introduction
Sugarcane is a major crop for sugar and bio-energy worldwide and essential for agricultural productivity and economic growth [1]. In China, sugar from sugarcane exceeds 90% of the total produced sugar; however, over 80% of the sugarcane planting area spans the slopes of arid lands in south China without irrigation facilities and where seasonal drought has become a major natural disaster. Seasonal drought is reflected in spatial-temporal characteristics based on the drought index of continuous days without precipitation [2]. Yield reduction caused by seasonal drought has gravely restricted the development of the sugarcane industry.
Several pot experiments and field studies have been conducted to understand the biological properties and improve the phenotypic and genotypic characteristics of sugarcane varieties under drought. Morphological parameters, including the leaf area, root and shoot weight, and stalk characteristics (length/height and diameter) of sugarcane, have been analyzed under drought induced via polyethylene glycol (PEG)treatment and withheld irrigation [3][4][5]. Both the drought-resistant and drought-susceptible genotypes showed decreased net photosynthetic rates (P n ), stomatal conductance (g s ), PSII quantum efficiency (Φ PSII ), and leaf relative water content; however, the genotypes exhibited increased nonphotochemical quenching (NPQ), malondialdehyde (MDA), proline, superoxide dismutase (SOD),and ascorbate peroxidase (APX) contents, which were higher in drought-resistant than in drought-susceptible genotypes [6][7][8]. Yield component analysis confirmed that the drought-resistant variety exhibited higher productivity than did the drought-susceptible variety due to its superior stalk characteristics (height, weight, and number) [9]. Moreover, studies on the roles of specific candidate genes (belonging to the abscisic-acid (ABA)-dependent pathway) in the drought-stress responses of resistant and susceptible sugarcane clones laid the foundation and direction for improving drought resistance of sugarcanes through marker-assisted breeding [10].
Process-based models have also been developed to estimate photosynthesis, biomass production, dry matter partitioning, and sugar accumulation for informed decision-making in sugarcane production [11,12]. Combining the irrigation module and the gross margin calculator with the crop growth and water balance model could simulate the yield and survival of newly planted ratoon sugarcane under different water-allocation and climate scenarios, supported with field irrigation management [13]. The accuracy of biomass simulation was substantially increased via further improving the transpiration efficiency and root-length density aspects of the model. These factors are determined throughthe genetic variations of sugarcane to water deficit; hourly transpiration traits, such as midday flattening and the conductance limit; and the hydraulic conductivity in each soil layer [14]. Moreover, the sugarcane model could achieve remote and large-scale monitoring of sugarcane irrigation when improved with innovative techniques and methods. These techniques involve adding a thermal infrared index coupled with a remote sense of the fraction of intercepted photosynthetically active radiation (fiPAR), or satellite thermal infrared [15][16][17]. However, almost all models are inseparable from simulation of photosynthetic performance, which is essential to plant growth as the source of matter and energy.
As a photosynthesis probe, chlorophyll fluorescence (ChlF) parameters have been applied to measure photosystem operation efficiency rapidly and nondestructively under normal and stressful circumstances [18]. The photosystem is directly affected by drought via stomatal closure triggered by ABA signals transferred from the roots [19,20], and thus could be used to evaluate the performances of plants subjected to drought stress. Decreases in the maximum quantum efficiency of PSII photochemistry (F v /F m ), Φ PSII , photochemical quenching (q P ), and the electron transport rate (ETR) accompanied increases in nonphotochemical quenching (NPQ), fast-relaxing NPQ (q E ), slowly relaxing NPQ (q I ), and the yield induced via downregulatory processes (Φ NPQ ); Φ NO had been shown under drought stress in many studies [21][22][23][24].
Among the chlorophyll fluorescence parameters, the reliability of F v /F m in evaluating the performance of sugarcane cultivars under drought stress was verified in a previous study [23]. The decreasing rate of F v /F m was significantly higher in drought-tolerance cultivars than that in drought-susceptible cultivars under drought stress [25]. Though F v /F m could become a stable drought-resistance indicator for sugarcane cultivars, considering its slow and slight decrease during drought stress (mostly 2~15%) [3,23,25], we chose to simulate another important parameter, Φ (namely Φ PSII ), under various water statuses, and evaluated sugarcane drought tolerance through analysis of the fitted parameters for two reasons: (1) Φ provides information on the noncyclic electron transport rate through PSII [18] and photoinhibition based on inactivation of PSII reaction centers [26] and exhibits a large varied amplitude (47.5%~64.3% decrease) with aggravation of drought in C 4 plants, including sugarcane and reed [24,27]; (2) Φ is directly related to the ETR, which provides energy to biochemical reactions of CO 2 assimilation [28], thus having been used to estimate photosynthetic rates via the modified FvCB model, which describes the relationship between the rate of carboxylation for the C 4 cycle (V p ) and the rate of ATP production driven via etransport (J ATP ), calculated from ∆F/F m (equal to Φ, where ∆F = F m − F s ,) [29]. Two other parameters, NPQ and Φ NO , were chosen in this study to analyze energy distribution under drought. The former represented total energy dissipation into heat loss containing energy-dependent, zeaxanthin-dependent, and photoinhibitory quenching [30], and the latter has been used to evaluate the occurrence of physiological damage accumulation during drought [24] and post-drought recovery [31].
Thus, the main objective of this study was to establish the relationship between photosynthetic quantum efficiency and water status and reveal the different droughtresponse patterns in drought-tolerant and -susceptible cultivars via analyzing the energy distribution associated with drought stress. This study provides a new perspective for sugarcane drought-tolerance assessment and drought-stress diagnosis.

Photothermal Model Parameterization and Its Biological Significance
Φ decreased with rising or falling temperature from T o under three given PAR levels, and the decreasing rate intensified with increasing PAR. The Φ bL values were lower than the Φ bH values under all three given PAR levels, indicating that the sugarcane photosystem is more sensitive to low temperatures. All of the Φ-related parameters (Φ bL , Φ bH , and Φ To ) declined with the increase in PAR, and the difference between Φ bL and Φ bH gradually increased, but both values occurred at the same PAR level in 'ROC22' and 'ROC16' (Figure 1). The F v /F m value showed no significant difference between 'ROC22' (0.7693 ± 0.003) and 'ROC16' (0.7685 ± 0.0027). Parameter E reflected the amplitude decrease in Φ To with increasing PAR and had the value of 5.91×10 −4 for both tested cultivars (Table 1). In general, the cultivars exhibited high light efficiency at low E. photosynthetic rates via the modified FvCB model, which describes the relationship between the rate of carboxylation for the C4 cycle (Vp) and the rate of ATP production driven via etransport (JATP), calculated from ΔF/Fm' (equal to Φ, where ΔF = Fm' − Fs,) [29]. Two other parameters, NPQ and ΦNO, were chosen in this study to analyze energy distribution under drought. The former represented total energy dissipation into heat loss containing energy-dependent, zeaxanthin-dependent, and photoinhibitory quenching [30], and the latter has been used to evaluate the occurrence of physiological damage accumulation during drought [24] and post-drought recovery [31].
Thus, the main objective of this study was to establish the relationship between photosynthetic quantum efficiency and water status and reveal the different drought-response patterns in drought-tolerant and -susceptible cultivars via analyzing the energy distribution associated with drought stress. This study provides a new perspective for sugarcane drought-tolerance assessment and drought-stress diagnosis.

Photothermal Model Parameterization and Its Biological Significance
Φ decreased with rising or falling temperature from To under three given PAR levels, and the decreasing rate intensified with increasing PAR. The ΦbL values were lower than the ΦbH values under all three given PAR levels, indicating that the sugarcane photosystem is more sensitive to low temperatures. All of the Φ-related parameters (ΦbL, ΦbH, and ΦTo) declined with the increase in PAR, and the difference between ΦbL and ΦbH gradually increased, but both values occurred at the same PAR level in 'ROC22' and 'ROC16' (Figure 1). The Fv/Fm value showed no significant difference between 'ROC22' (0.7693 ± 0.003) and 'ROC16' (0.7685 ± 0.0027). Parameter Ε reflected the amplitude decrease in ΦTo with increasing PAR and had the value of 5.91×10 −4 for both tested cultivars (Table 1). In general, the cultivars exhibited high light efficiency at low Ε.

Response of the Estimated Parameter to Drought Treatment
The parameter E D increased with the decrease in rSWC for both cultivars, and a higher E D value of denoted severe effects of drought-induced stress on the plant thus could be used for drought-stress assessment. Different downward trends were observed between the two cultivars during drought stress. The E D of 'ROC22' increased gradually when the rSWC decreased from 29% to 18%, while that of 'ROC16' increased drastically when the rSWC dropped below the 29% level. Based on the corresponding E D for droughtstress assessment, the slight, moderate, and severe drought conditions were described as 7 <E D < 9, 9 <E D < 13, and E D > 13, respectively ( Figure 2a).
According to our experiments, the rSWC c values for 'ROC22' and 'ROC16' were 40% and 29%, respectively, showing that the photochemical reaction for 'ROC22' responded at a higher water status in contrast to that of 'ROC16' (Figure 2b). Additionally, rSWC decreased faster for 'ROC16' than for 'ROC22' during drought stress, indicating that the two cultivars possessed different water consumption characteristics under drought conditions. could be used for drought-stress assessment. Different downward trends were observed between the two cultivars during drought stress. The ΕD of 'ROC22' increased gradually when the rSWC decreased from 29% to 18%, while that of 'ROC16' increased drastically when the rSWC dropped below the 29% level. Based on the corresponding Ε D for drought-stress assessment, the slight, moderate, and severe drought conditions were described as 7 <Ε D< 9, 9 <Ε D< 13, and Ε D> 13, respectively (Figure 2a). According to our experiments, the rSWCc values for 'ROC22' and 'ROC16' were 40% and 29%, respectively, showing that the photochemical reaction for 'ROC22' responded at a higher water status in contrast to that of 'ROC16' (Figure 2b). Additionally, rSWC decreased faster for 'ROC16' than for 'ROC22' during drought stress, indicating that the two cultivars possessed different water consumption characteristics under drought conditions.

Energy Distributionin PSII in Response to Drought
We used the parameters Φ and NPQ to assess the transfer of excitation light energy to photochemical reaction and its decay via heat loss, respectively. Five days after the drought treatment, the Φ of 'ROC16' was slightly higher than that of 'ROC22'; however, both cultivars exhibited the same NPQ level on day 1, and it differed with the increasing PAR from day 3 to day 5 (Figure 3a-c). 'ROC22' maintained high Φ compared to 'ROC16' at day 7 after the drought treatment, but the two cultivars had the same NPQ level when the PAR was below 900 μmol photons m −2 s −1 . After 9 days of drought stress, both the Φ and the NPQ of 'ROC16' were significantly lower than those of 'ROC22' (Figure 3d,e). The results demonstrated that the photoprotective mechanism based on the heat loss of 'ROC16' did not involve excessive energy consumptionwhen compared with that of 'ROC22'. Another piece of evidence that showed less energy consumption was the increased ΦNO levels observed on day 3 (rSWC 56%) and day 5 (rSWC 40%) for 'ROC16' and 'ROC22', respectively, after the drought treatment. The ΦNO value of 'ROC16' was significantly higher than that of 'ROC22' after day 3, and the difference was magnified with aggravation of drought stress, indicating higher drought-induced damage imposed on 'ROC16' than on 'ROC22' (Figure 3f).

Energy Distributionin PSII in Response to Drought
We used the parameters Φ and NPQ to assess the transfer of excitation light energy to photochemical reaction and its decay via heat loss, respectively. Five days after the drought treatment, the Φ of 'ROC16' was slightly higher than that of 'ROC22'; however, both cultivars exhibited the same NPQ level on day 1, and it differed with the increasing PAR from day 3 to day 5 (Figure 3a-c). 'ROC22' maintained high Φ compared to 'ROC16' at day 7 after the drought treatment, but the two cultivars had the same NPQ level when the PAR was below 900 µmol photons m −2 s −1 . After 9 days of drought stress, both the Φ and the NPQ of 'ROC16' were significantly lower than those of 'ROC22' (Figure 3d,e). The results demonstrated that the photoprotective mechanism based on the heat loss of 'ROC16' did not involve excessive energy consumptionwhen compared with that of 'ROC22'. Another piece of evidence that showed less energy consumption was the increased Φ NO levels observed on day 3 (rSWC 56%) and day 5 (rSWC 40%) for 'ROC16' and 'ROC22', respectively, after the drought treatment. The Φ NO value of 'ROC16' was significantly higher than that of 'ROC22' after day 3, and the difference was magnified with aggravation of drought stress, indicating higher drought-induced damage imposed on 'ROC16' than on 'ROC22' (Figure 3f). Furthermore, the E D values of the two cultivars increased on day 5 after the drought treatment, but the rSWC values of 'ROC22' and 'ROC16' decreased to 40% and 29%, respectively. This differed from the rSWC values (56% for 'ROC16' and 40% for 'ROC22') that corresponded to the turning point of Φ NO , implying that the PSII of 'ROC22' responded to the water deficit more quickly than did that of 'ROC16'. Contemporaneously, for 'ROC22', the NPQ rose to 1.6 when the PAR approached 1200 µmol photons m −2 s −1 but remained below 1.3 for 'ROC16' throughout the drought treatment. NPQ merely increased when the PAR ranged from 100 to 900 µmol photons m −2 s −1 on day 7 and day 9 for 'ROC16' and on day 5 for 'ROC22'. This phenomenon demonstrated that rapid and high energy dissipation in drought is essential for drought tolerance in sugarcane.

Different Drought-Response Patterns for Two Cultivars
We investigated the diverse drought-response patterns of the two sugarcane cultivars. The results showed that 'ROC22' exhibited a quick response and a high capability of NPQ (Figure 2b) instead of liner electron transport (exhibited via low E D ) at the onset of drought (rSWC of 40%) (Figure 3c). This delayed the conversion from the drought-induced reaction to the physiological damage of the PSII, which was indicated with an increase in Φ NO (Figure 3f). However, 'ROC16' maintained a relatively lower E D value to ensure photochemical efficiency until the rSWC dropped to 29%, after which the NPQ increased at PAR of merely 100~900 µmol photons m −2 s −1 . Moreover, the Φ NO of 'ROC16' was significantly higher than that of 'ROC22'. The different water consumption characteristics of the two cultivars were revealed in the differentially decreasing rates of rSWC, which showed that 'ROC16' had higher rSWC than 'ROC22' throughout the drought treatment.

Model Validation
The coefficient of determination (r 2 ) and relative root mean-squared error (rRMSE) of the predicted values were 0.922 and 0.11 (Figure 4a), while those of the measured values were 0.826 and 0.309 (Figure 4b), respectively, for Φ under different photothermal and drought conditions, indicating that the model could be applied for drought-tolerance assessment or drought-stress diagnosis for sugarcane cultivars. characteristics of the two cultivars were revealed in the differentially decreasing rates of rSWC, which showed that 'ROC16' had higher rSWC than 'ROC22' throughout the drought treatment.

Model Validation
The coefficient of determination (r 2 ) and relative root mean-squared error (rRMSE) of the predicted values were 0.922 and 0.11 (Figure 4a), while those of the measured values were 0.826 and 0.309 (Figure 4b), respectively, for Φ under different photothermal and drought conditions, indicating that the model could be applied for drought-tolerance assessment or drought-stress diagnosis for sugarcane cultivars.

Discussion
The minimum, optimal, and maximum temperatures obtained from the curve fitting of the net photosynthetic rate under saturated light (Pn,max) and air temperature (Ta) were used to describe the fundamental temperature for crop growth [32]. The relationship between Φ and T at limited PAR observed in our study was similar to those reported in previous studies [33]; however, in our study, the fitted parameters varied with different PAR levels ( Figure 1). Unlike with Pn,max, the occurrence of Φb (ΦbL or ΦbH) at low or high temperatures was associated with alternative electron fluxes beyond photochemical reaction processes, such as photorespiration, the Mehler reaction, or cyclic and pseudocyclic electron transport under normal or drought conditions [34,35]. For instance, photorespiration and the Mehler reaction constituted 20% and 30% of the total electron flux, respectively [36,37]. Moreover, the rETR remained at 50% under normal conditions, even when photosynthesis ceased due to stomatal closure under drought stress [38].This could also explain the existence of Φb and the gradual decrease

Discussion
The minimum, optimal, and maximum temperatures obtained from the curve fitting of the net photosynthetic rate under saturated light (P n,max ) and air temperature (Ta) were used to describe the fundamental temperature for crop growth [32]. The relationship between Φ and T at limited PAR observed in our study was similar to those reported in previous studies [33]; however, in our study, the fitted parameters varied with different PAR levels ( Figure 1). Unlike with P n,max , the occurrence of Φ b (Φ bL or Φ bH ) at low or high temperatures was associated with alternative electron fluxes beyond photochemical reaction processes, such as photorespiration, the Mehler reaction, or cyclic and pseudocyclic electron transport under normal or drought conditions [34,35]. For instance, photorespiration and the Mehler reaction constituted 20% and 30% of the total electron flux, respectively [36,37]. Moreover, the rETR remained at 50% under normal conditions, even when photosynthesis ceased due to stomatal closure under drought stress [38]. This could also explain the existence of Φ b and the gradual decrease in Φ with increases or decreases in temperature in our study compared with the P n,max -Ta of the C 3 , C 4, and CAM plants [39].
The Φ of sugarcane, a typical C 4 crop, showed higher sensitivity to lower temperatures than to higher temperatures because C 4 pathway enzymes are cold-labile due to the limitations of phosphoenolpyruvate carboxylase (PEPC) and pyruvate phosphate dikinase (PPDK) [40,41]. This explains why the Φ bL value was lower than that of Φ bH , and the increasing disparity between Φ bL and Φ bH with increasing PAR. The assumption that Φ bL and Φ bH exhibited the same downward trend as Φ To in our study resulted in underestimation of Φ, especially when a value below 0.2 was obtained with a combination of drought, high PAR, and low (or high) temperature (Figure 4b). This phenomenon resulted from the fact that both net photosynthetic rate and Φ PSII significantly decrease more under drought-cold stress than under drought stress only [42]. Additionally, combined heat and drought stress severely affect crop leaves at the physiological and biochemical levels via detrimental variation of photosynthetic pigments, osmolytes, and enzymatic antioxidant activities more than heat or drought stress alone [43,44].
We supposed that light energy was allocated between liner electron transport (Φ) and heat loss (NPQ) at the onset of drought stress until the occurrence of potential damage in the PSII, indicated with the concurrence of increased Φ NO and decreased NPQ (or q N ) shown in studies of salt stress [45,46] and disease influence [47]. The increase in NPQ compromised the decrease in Φ based on their new relationship acquired in the dark that followed illumination, which could be more appropriate to describe the photoprotective potential of NPQ [48]. This phenomenon also appeared in sweet sorghum and maize, with the degree of NPQ change differing in different cultivars under drought stress [49,50]. However, a significant increase in Φ NO was found when sorghum suffered from drought [51]; this might be related to quenching of singlet oxygen via β-carotene [52], which could also cause photoinhibition via degradation of the D1 protein in the PSII center [53]. Similar results appeared in coastal halophytic marsh grasses with increasing salinity, and the high levels of the Mehler reaction and antioxidant enzymes improved adaptation in photoprotection [54]. Another study reported a decrease in Φ NO with enhanced Φ NPQ in maize under a long time field condition, which probably represented heat energy loss in protection of the stressed younger leaves [55].
Both the Φ and the water consumption of 'ROC22' were lower than those of 'ROC16' before the 5-day marker after drought, similarly to a previous study, where the sugarcane response to drought involved a rapid decrease in P n and g s for the drought-tolerance cultivar rather than the drought-sensitive one [7]. This might be associated with rapid stomatal closure, to reduce water loss, induced via ABA [56], which is synthesized in arid root systems to induce specific genes and proteins in response to water deficit [10,57]. In conversion of the value range of the E D to water status, the drought degree of sugarcane plantation was divided into three stages: 29% < rSWC < 40% (slight), 18 < rSWC< 29% (moderate), and rSWC < 18% (severe). These ranges were lower than those reported in the previous study (which were 65~70% (slight), 45~50% (moderate), and 25~30% (severe)) [58], since we concentrated on the critical value of irrecoverable physiological injury instead of the yield reduction caused by the stress reaction. The accuracy, persuasiveness, and application scope of the model could be improved after more cultivars (or varieties) participate in parameter fitting and model validation.

Plant Materials and Treatment
Two sugarcane cultivars, 'ROC22' (drought-resistant) and 'ROC16' (drought-sensitive), cultivated by the Taiwan Sugar Research Institute, were used in this study; both were main cultivars introduced and applied in production in south China [59] and used in many drought-response researches [24,60,61]. Five experiments were conducted with the same planting, management, and drought treatment during different seasons in Zhanjiang, China (21 • N, 110 • E), from 2019 to 2021. Bucket planting was adopted for the experiment, with 30 buckets for each cultivar and three stems (1 bud per segment) per bucket, arranged based on field planting density (90,000 buds·ha −1 ). The caliber, bottom diameter, and height of each bucket were 40 cm, 30 cm, and 40 cm, respectively. The buckets contained red soil (70%), sand (20%), and organic manure (10%) as a substrate. The buckets were placed on hardened ground in an open environment without shading to implement photothermal experiments at the 8 th /euphylla stage until the drought experiments were initiated. Five and ten buckets with uniform-growth seedlings were selected for the photothermal experiments and the drought experiments, respectively. The environmental conditions and substrate properties for each experiment are displayed in Table 2. Note: a and b denote photothermal and drought experiments, while M and V denote experimental data used for modeling and validation, respectively. Temperature and PAR data were selected from the daily data set recorded with a farm land environment monitoring system (CR1000).
Photothermal experiments were conducted under field conditions. The average F v /F m value of 9 leaves, measured at T o under well-watered conditions and selected from 15 plants (5 buckets) in experiment 5 (Exp5) and with 30 points of Φ obtained under 3 given PAR levels (285, 625, and 1150 µmol photons m −2 s −1 ), with T ranging from 11.8 • C to 42.7 • C for each cultivar from 30 plants (10 buckets) in Exp1-3, was used to fit the photothermal response model (evaluation of Φ under different light and temperature backgrounds). Furthermore, 60 sets of the Φ, PAR, and T measured for each cultivar fromExp4 and Exp5were used for model validation. The buckets were kept in a nearby greenhouse for a short time (10 min) for Φ acquisition at a higher T (over 38 • C).
For the subsequent drought experiments, the buckets were placed infield conditions, except for those in Exp5, which were moved into the phytotron from 6:00 to 12:00, with the PAR increasing from 0 to 1200 µmol photons m −2 s −1 (20% light intensity increased per hour using artificial light) under optimal T (29 ± 1.5 • C) at 1, 3, 5, 7, and 9 d after withholding of irrigation. The measured Φ, VWC, F v /F m , PAR, and T were used to estimate E D , and the energy dissipation parameters obtained from Exp5 were used to analyze the different responses of the photoprotective mechanisms between the two cultivars under drought conditions. A total of 125 sets of the Φ, VWC, F v /F m , PAR, and T obtainedfor each cultivarfrom 8:00 to 12:00 am in Exp1-4 were used for model validation.

Chlorophyll Fluorescence and Substrate Water Content Measurement
The chlorophyll fluorescence (ChlF) parameters were measured on the middle part of each +1 leaf (the first fully expanded leaf when counting from the top down) using a MINI-PAM-II analyzer (Walz, Effeltrich, Germany), which measures PAR and temperature via a self-contained sensor. All of the ChlF parameters were obtained under natural light conditions, except for those measured in the phytotron in Exp5, where artificial light had been used for light adaptation. The leaf clip was slightly adjusted to ensure relatively constant PAR (within ±10), and a stable F s was waited for to initiate the process of measuring F m . Then, F o and F m were measured after 30 min of dark adaptation using dark clips at the same leaf position. The next round of F s and F m measurement was conducted after 30 min of natural light adaptation (artificial light in the phytotron in Exp5). Four basic fluorescence parameters, F s (steady fluorescence from light-adapted leaves), F m (maximal fluorescence from light-adapted leaves), F o (minimal fluorescence from 30 min dark-adapted leaves), and F m (maximal fluorescence from 30 min dark-adapted leaves), were measured from morning to midday. The obtained parameters were then used to calculate the minimal fluorescence from light-adapted leaves (F o ) [62], photosynthetic quantum efficiency (Φ), the maximum quantum efficiency of PSII photochemistry (F v /F m ), nonphotochemical quenching (NPQ), and the yield for other energy losses (Φ NO ) [63] indicating non-regulated energy dissipation.
Volumetric water content (VWC) was measured at a depth of 0~16 cm in each bucket via AZS-100 (Aozuo, China) at 8:00 a.m., with 5 repetitions during the drought treatment. The rSWC was calculated with the formula rSWC = VWC/VWC S (VWC S was measured at sufficient irrigation with 3 repetitions; Table 1).

Estimation of Photosynthetic Quantum Efficiency (Φ) under Different Photothermal Conditions
Three levels of PAR (285, 625, and 1150 µmol photons m −2 s −1 ) were adopted to measure Φ under different air temperatures (T) using artificial light in field conditions in different seasons, calculated using the formulae transformed from the calibration model proposed to evaluate the response of Φ to photothermal backgrounds [64] (Figure 1a): where Φ To , Φ bL , and Φ bH represent the optimum Φ under T o and the basic Φ values under low T and high T at limited PAR. T min , T o , and T max represent the minimal, optimal, and maximal temperatures for sugarcane photosynthesis and vegetative growth, with values of 15 • C, 30 • C, and 40 • C, respectively, according to previous studies [65][66][67]. The relationship between Φ bL , Φ bH , Φ To , and PAR was established as follows ( Figure 1b): The mean value of the F v /F m measured from Exp1~3 was used for Equation (4)'s fitting, and a L , b L , a H , and b H are empirical coefficients fitted from Equations (2) and (3). The parameter E, fitted from Equation (4), was merely related to different sugarcane cultivars.

Estimation of the Parameters under Drought Conditions
We inserted Equations (2)-(4) into Equation (1) to determine the Φ, F v /F m , PAR, T, and rSWC measured under drought and used the values to describe the relationship between ε and rSWC (Figure 2a): where rSWC c indicates the critical limit of rSWC for the two sugarcane cultivars, E D is E under drought stress, and c and d are the empirical coefficients fitted from Equation (5).
The effect of drought stress on Φ bL and Φ bH was assumed to be the same as that of Φ To , and these values were calculated as: where Φ bLD and Φ bHD represent Φ bL and Φ bH , respectively, under drought stress.

Model Validation
The coefficient of determination (r 2 ) and relative root mean-squared error (rRMSE) were adopted to analyze the conformity and accuracy between the predicted and the measured values and were calculated as follows: where x, y, x, y, OBS i , SIM i , M, and n are the measured value, predicted value, average measured variables, average predicted variables, observed value, simulated value, average observed value, and sample size for Equations (8) and (9), respectively. The definitions, units, and values (standard error shown in brackets) of the parameters and empirical coefficients are listed in Table 1.

Conclusions
The simulation and energy distribution analyses exhibited an opposition between water consumption and drought tolerance. 'ROC22' (drought-resistant) decreased its water consumption and activated a photoprotective mechanism to delay the droughtinduced physiological injury, while 'ROC16' (drought-susceptible) maintained relatively high water consumption to guarantee photochemical reactions but was severely injured with aggravation of drought.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding authors.