Effect of Elevated Air Temperature on the Growth and Yield of Paddy Rice

Abstract

Rice is one of the major food crops, particularly in Asia. However, it is vulnerable to high temperature and has high yield fluctuations. Monitoring crop growth and physiological responses to high temperatures can help us better understand the agricultural impacts of global warming. The aim of this study is to monitor growth, development, and physiological responses to high temperature conditions on paddy rice and to assess their combined effects on yield. In this study, changes to growth, maturity, and senescence in paddy rice throughout the growing season were identified under elevated air temperature conditions created by a temperature gradient field chamber (TGFC). That facility provides a gradient from the ambient air temperature (AT) to 3 °C above AT (AT + 3 °C). To represent crop physiology and productivity, we measured the plant height, chlorophyll, normalized difference vegetation index (NDVI), and maximum photosynthetic rate (A_max) to assess growth and physiological processes, and heat stress effects on four yield measurements were assessed using the heating degree day index. Rice height increased more rapidly in the AT + 3 °C treatment from the early growth stage to heading, while SPAD and NDVI decreased more rapidly at AT after heading. The A_max of AT and AT + 3 °C was not significantly different in the tillering stage. However, it was higher at AT in the booting stage but higher at AT + 3 °C in the grain filling stage. These results indicate that paddy rice was not affected by heat stress at the tillering stage, but a cumulative effect emerged by the booting stage. Further, photosynthetic capacity was maintained much later into the grain filling stage at AT + 3 °C. These results will be useful for understanding the growth and physiological responses of paddy rice to global warming.

1. Introduction

Rice is a staple crop for more than half the world’s population, and the cultivation and consumption of rice are widespread in the Asian region [1]. However, it is vulnerable to high temperature and has high yield fluctuations [2]. The impacts of global warming, associated with climate change, are increasing over time. According to the IPCC’s 2021 special report, if the current trend continues, the global mean temperature increase will exceed 1.5 °C above pre-industrial levels between 2030 and 2052, and is projected to reach 3 °C by 2100 [3]. In addition, abnormal weather phenomena, such as extreme high temperatures, are expected to occur more frequently in most areas. Recently, abnormal high temperature phenomena, including heatwaves and tropical nights, have frequently occurred in the Korean peninsula [4,5]. Rice is one of the three major food crops and the second largest food supply in the world after wheat, meeting about 80% of the food calorie requirements of more than half of the world’s population. It is a critical crop in the context of food security, but it is vulnerable to high temperatures and has high yield fluctuations. In particular, high temperatures during the reproductive and flowering stages are directly linked to the yield [6,7,8,9], and by 2050, 27% of rice-growing areas are expected to experience at least five days of heat stress during these stages [10].

Previous studies have shown that heat stress shortens the growth and development stage of crops (e.g., [11,12,13]). Moreover, extremely high temperatures reduce the size of crops and decrease the total amount of photosynthesis during the whole growing season [14]. Kim et al. [15] reported that the early termination of grain filling in high temperature conditions was due to a loss of sink capacity rather than leaf senescence, resulting in an increased distribution of dry matter to leaves and stems.

Crop heat stress due to heatwave events will affect growth and physiological processes, ultimately expressed as yield and biomass losses [11]. In particular, photosynthesis and respiration, which are sensitive to temperature conditions, can affect the total biomass of a crop and are often responsible for final yield fluctuations [16]. Many studies have shown that the photosynthetic response to temperature can be explained as a parabolic curve in which photosynthesis is suppressed at low and high temperatures, but the optimal temperature also changes seasonally, with mutable photosynthetic efficiency at varying temperatures [17]. However, it has been reported that light use efficiency at the leaf and canopy scale decreases under stress conditions, such as high and low temperatures, drought, and biotic stresses [18,19].

Increased mean air temperature and abnormal high temperature events, such as extreme heatwaves, affect the growth and physiological conditions of crops throughout the growing season. Moreover, high temperature phenomena during the growing season reduce a crop’s capacity to gain any possible beneficial effects of increased accumulated degree days under global warming [20]. The benefits of the expected rise in CO₂ concentrations can also be eliminated due to temperature increases [21]. In addition, even if the temperature is not high throughout the growing period, extreme events such as heatwaves can cause serious damage to rice production because pollen viability in rice exposed to high temperatures is reduced [6,22]. This spikelet sterility will not only reduce yield but will also affect growth and physiological responses during the reproductive growth period.

Monitoring and quantifying the growth and physiological responses of crops to global warming is critical for making informed policy decisions that affect food security. Thus, non-destructive and efficient optical measurement methods have been used in many studies to assess the impacts of climate change on crops. In particular, the SPAD value, which is clearly proportional to the concentration of nitrogen present in the leaves, has been used in many studies to determine the chlorophyll content in the leaves. In addition, many studies have demonstrated that photosynthetic capacity and leaf growth vary with leaf nitrogen content [23,24,25]. The normalized difference vegetation index (NDVI) is widely used to represent the biomass and health of vegetation, and it is useful for monitoring the growth of crops; moreover, the accumulated effect of the environment on a crop can be expressed as photosynthesis at the leaf level. Therefore, the purpose of this study is to monitor growth, development, and physiological responses to high temperature conditions on paddy rice and to assess their combined effects on yield.

2. Materials and Methods

2.1. Experimental Facility and Conditions

This study was conducted in a sunlit temperature gradient field chamber (TGFC) installed in a paddy field at the Agricultural Practice Training Center of Chonnam National University, Gwangju, Korea (35°10′ N, 126°53′ E; elevation 33 m) in 2016 and 2018. The TGFC is a special-purpose heating facility based on field conditions that consists of one heater and three ventilators: two small fans and one large fan. The entrance is always opened, and the innermost ventilator operates to make the air moving from the entrance to the back smoothly through the chamber. Along the length of the TGFC, four units, labeled T0, T1, T2, and T3, are installed to measure the air temperature and humidity from the entrance (T0) to the back of the chamber (T3). Using the heater and ventilators, the TGFC is designed to maintain the temperature gradually increasing to a 3 °C higher air temperature (AT + 3 °C) at the innermost part of the chamber, compared to the ambient temperature (AT) at the entrance. When the air temperature difference between T0 (AT) and T3 (AT + 3 °C) is lower than 3 °C, the heater is operated to inject heated air, and when it is higher than 3 °C, the speed of the ventilator fans is increased [26]. A detailed depiction of the TGFC is shown in Figure 1.

The daily mean, maximum, and minimum air temperatures during the rice growing season (from transplanting to 120 days after transplanting) were 1.2, 1.1, and 1.3 °C greater in 2016 and 1.5, 2.1, and 1.3 °C greater in 2018 relative to the average of 1981–2010 (Figure 2). In particular, there was an unprecedented heatwave event in July and August 2018. In this study, the Ilmi variety of Oryza sativa L. japonica, one of the leading Korean rice cultivars, was used because it is known to be resistant to high temperature stress [27]. The mid-late maturing Ilmi cultivar was transplanted after sowing the seeds into the TGFC on 3 June 2016 (DOY 155), and 1 June 2018 (DOY 152), and harvested on 13 October 2016 (DOY 287), and 27 September 2018 (DOY 270). The number of transplanted rice plants was 3 per hill, and the plant density was 15 × 30 cm. In this experiment, three TGFCs were used in both years, with 960 (12 × 80) rice plants grown per TGFC in an area of 43.2 m² (1.8 × 24 m). Fertilizer treatments of 845.2 g N (nitrogen), 972 g P (phosphorus), and 547.2 g K (potassium) per TGFC, based on an N:P:K of 9:4.5:5.7 kg per 10 acre, were applied at planting. Two more supplemental fertilizers were applied based on a 5:2:3 ratio.

2.2. Meteorological Data Acquisition

Air temperature (T_a) and relative humidity (RH) were measured at T0, T1, T2, and T3, from the entrance to the back of the TGFC. The data were recorded every 5 min in a data logger (CR10X; Campbell Scientific, Logan, UT, USA) and used to automatically maintain the temperature gradient in the TGFC. Other meteorological data, including photosynthetically active radiation (PAR) and solar radiation, were measured inside and outside the TGFC using a quantum sensor (LI-190R; LiCor Inc., Lincoln, NE, USA) and a pyranometer (LI-200R; LiCor Inc., Lincoln, NE, USA), respectively. Temperature data for the experimental years 2016 and 2018, and for the years 1981–2010, in order to contextualize the relative weather conditions of the experimental years, were obtained from the Korea Meteorological Administration (KMA, https://data.kma.go.kr/cmmn/main.do (accessed on 21 November 2023)).

2.3. Heating Degree Day (HDD) Index

The optimum and threshold growth temperature varies according to cultivars and developmental stages, with limits to how much this can be simplified [28,29]. However, in general, the optimum growth temperature of rice is known to be from 20 to 35 °C, and the threshold temperature affecting grain yield is 34–35 °C [6,7,28,30]. The temperature affects various determinants of rice yield and contributes to variation in overall yield [31]. The number of panicles per unit area is directly linked to the number of effective tillers and depends on environmental conditions post transplanting, but it is almost unaffected after the spikelet differentiation stage. The number of spikelets per panicle is determined by the difference between the number of differentiated spikelets and the number of spikelets that degenerate after forming. The differentiation and degeneration of spikelets are most likely to be affected by the environment for 5 to 35 days before heading [32]. The grain filling rate and 1000-grain weight depend on environmental conditions after the heading stage. Heat stress during this period causes spikelet sterility and reduces the accumulation of starch and amylose, negatively affecting the grain filling rate and quality [33].

According to previous studies, the final grain yield shows the most significant relationship with the grain filling rate, and the higher the intensity of heat stress at anthesis, the more the sterility increases, and the more conspicuous the relationship becomes. Also, since the harvest index (HI) depends on the final grain yield, it may change heavily depending on the environmental conditions after the heading stage [29]. Heat stress in plants is a complex function of the intensity and duration of high temperatures. Additionally, in the evaluation of heat stress in crops, not only the instantaneous temperature of the moment but also the accumulated temperature is important. Accumulated heat stress is expressed as growth effects through physiological processes, ultimately affecting the final yield. Thus, in this study, the heating degree day index (HDD) was used as a stress index to comprehensively evaluate the duration and intensity of heat stress. The HDD was calculated for each temperature regime using a modified version of the equations described by Shi et al. [34]:

$${\mathrm{H}\mathrm{D}\mathrm{D}}_{{35}_{t+5}^{t+50}}={\sum }_{d_t+5}^{d_t+50}{\mathrm{H}\mathrm{D}}_i$$

(1)

$${\mathrm{H}\mathrm{D}\mathrm{D}}_{{35}_{h-35}^{h-5}}={\sum }_{d_h-35}^{d_h-5}{\mathrm{H}\mathrm{D}}_i$$

(2)

$${\mathrm{H}\mathrm{D}\mathrm{D}}_{{35}_h^m}={\sum }_{d_h}^{d_m}{\mathrm{H}\mathrm{D}}_i$$

(3)

$${\mathrm{H}\mathrm{D}}_i=\frac{1}{24}{\sum }_{j=1}^{24}{\mathrm{H}\mathrm{D}\mathrm{D}}_j$$

(4)

$${\mathrm{H}\mathrm{D}\mathrm{D}}_j=\left\{\begin{array}{c}0\\ T_j-T_h\end{array}\right.\genfrac{}{}{0pt}{}{T_j\le T_h}{T_j>T_h}\left(\mathrm{j}=1,2,\dots ,24\right)$$

(5)

where d_t, d_h, and dm are the date of transplanting, heading, and maturity; T_j is the hourly temperature, which is based on the temperature recorded every 5 min; and T_h is the threshold temperature, which was set at 35 °C. Equations (1)–(3) represent the HDD for specific factors related to grain yield over a specified period: Equation (1) is for the number of panicles per unit area; (2) is for the number of spikelets per panicle; and (3) is for the yield components, total grain yield, and HI. Equations (4) and (5), respectively, quantify and sum the temperature values above the threshold temperature on day i.

2.4. Monitoring of Plant Growth

To identify growth responses and development stages, the plant height and number of tillers were investigated in 10 plants in each temperature regime once a week starting 2 weeks after transplanting. The SPAD value was also measured once a week after the heading stage (69, 75, 84, 89, 96, 110, and 119 days after transplanting (DAT) in 2016 and 75, 83, 90, 97, 105, 112, and 119 DAT in 2018) to monitor changes in the chlorophyll contents of leaves at the leaf level using a portable chlorophyll meter (SPAD-502; Minolta Corp., Osaka, Japan) based on the differential transmission of red and nearinfrared (NIR) radiation. A high SPAD value indicates a large amount of chlorophyll in the leaves. For the SPAD measurement, 20 of the topmost fully expanded leaves were randomly selected, and the average of consecutive values measured 10–15 times at 5 cm away from the leaf tip to the middle of the leaf were used.

A multispectral radiometer (MSR16R; CROPSCAN Inc., Rochester, MN, USA) and spectrometer (AvaSpec-ULS-2048L; Avantes, Apeldoorn, The Netherlands) were used to measure the reflectance of paddy rice in 2016 and 2018, respectively. A multispectral radiometer has 16 wavebands in the 450–1750 nm region and a field of view (FOV) of 28 degrees. The 660 and 800 nm wavelengths were used in this study. A spectrometer has wavebands in the 300–1100 nm range and a 23-degree FOV. All measurements of reflectance were conducted at midday under clear-sky conditions at intervals of 7 to 14 days after the heading stage (69, 78, 82, 89, 97, 109, and 123 DAT in 2016 and 73, 77, 89, 95, 100, 109, and 117 DAT in 2018). The NDVI is useful for monitoring the growth of crops. It was calculated as [35]:

$$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}=\frac{(R_{800}-R_{660})}{(R_{800}+{ R}_{660})}$$

(6)

where R₈₀₀ is the reflectance at 800 nm and R₆₆₀ is the reflectance at 660 nm. In this study, interpolated NDVI data were used to monitor time-series variations in the chlorophyll status of leaves, which were related to the photosynthetic capacity at the canopy level.

2.5. Leaf Gas Exchange Measurements

To determine the photosynthetic capacity of leaves at three main phenological stages—tillering (DAT 38 and 41), booting (DAT 41 and 52), and late grain filling (DAT 90) in 2016 and tillering (DAT 27), booting (DAT 53), and late grain filling (DAT 97) in 2018—under two air temperature regimes of AT and AT + 3 °C, light response curves were measured using an LI-6400 portable photosynthesis system incorporating an infrared gas analyzer (LiCor Inc., Lincoln, NE, USA) and equipped with a 2 × 3 cm opaque leaf cuvette with an internal red-blue LED light source (6400-02B LED; LiCor Inc., Lincoln, NE, USA). In addition, as supplementary data, chlorophyll contents were measured for all samples using a portable chlorophyll meter. As samples for the measurement of leaf-level gas exchange, 3 typical middle to topmost fully expanded leaves were selected. All measurement data were recalculated using the actual leaf area in the interior area (6 cm²) of the leaf cuvette.

All measurements were performed at varying photosynthetically active radiation levels between 09:00 and 12:00 on clear days. Also, in order to reduce the stress to the leaf, each measurement was carried out after adaptation to the environmental conditions inside the leaf cuvette for at least 10–15 min. The measurement entailed starting from the light intensity equivalent to ambient light and measuring from saturated light to weak light, with transitions between light intensity made after the intercellular CO₂ concentration and net photosynthetic rate remained constant for at least 5 min. The conditions inside the leaf cuvette were controlled at a leaf temperature of 25–30 °C and relative humidity (RH) of 40–70%, and a fixed reference CO₂ concentration of 400 μmol mol⁻¹ was maintained using a desiccant, soda lime, and 12 g CO₂ cylinders (LiCor Inc., Lincoln, NE, USA). A maximum photosynthetic rate (A_max) was calculated for each light response curve using the A/Q curve protocol of the LI-6400 software (https://www.licor.com/env/support/LI-6400/software.html (accessed on 21 November 2023), LiCor Inc., Lincoln, NE, USA). The light-use efficiency at the leaf level (LUE_leaf) was calculated as the slope of the linear regression of PAR between 0 and 400 or 500 μmol m⁻² s⁻¹.

2.6. Yield and Biomass

All yield components and the biomass were measured after harvest. Samples were selected as 10 plants in the center of the experimental plot in each temperature regime, considering border effects. The harvested samples were brought into the laboratory and immediately separated into stem, leaf, and panicles, and then they were dried in a drying oven at 70 °C for 3 days or 80 °C for 2 days, and the above-ground dry matter (AGDM) was measured. Grain yield was estimated using four yield components: the number of panicles per m², the number of spikelets per panicle, the grain filling rate (%), and the 1000-grain weight (g). The number of panicles per m² was estimated by multiplying the average number of panicles per hill and the number of hills per m². HI was calculated by dividing the filled grain weight by the AGDM.

2.7. Statistical Analysis

The growth, physiological responses, and yield of paddy rice to elevated air temperature were statistically analyzed for two temperature regimes, AT and AT + 3 °C. Independent two-sample t-tests were conducted to verify the significant differences between temperature regimes for growth and photosynthetic parameters. An analysis of variance (ANOVA) was carried out to assess the statistical significance of the difference of the year and temperature regime and their interaction on yield components, total yield measures, and AGDM. Regression analyses relating yield components and total yield measures to HDD were conducted to characterize the relationships and determine thresholds. All statistical analyses were performed with SPSS (IBM SPSS Statistic for Windows, v23.0, IBM Corp., Armonk, NY, USA).

3. Results and Discussion

3.1. Growth Responses to Elevated Air Temperature

The growth responses to elevated air temperature over the whole growing season were divided into two parts: plant height was observed to compare the physical size changes of rice from the early growth stage to heading. After heading, the changes in SPAD value, indicating the chlorophyll content of leaves [25], and NDVI, indicating the amount of vegetation, greenness, and photosynthetic capacity of vegetation at the canopy level, were also monitored (Figure 3). The plant height during the vegetative growth stage showed a increase under both temperature regimes. Specifically, the plant height increased rapidly in the effective tillering stage but slowed down in the non-productive tillering stage. The plant height then increased again in the reproductive stage, including during panicle initiation. However, a significant resumption of growth during the reproductive stage in the AT + 3 °C treatment did not occur in 2016, with only a slight increase before heading. Some previous studies (e.g., [14,29]) reported that elevated temperature conditions can often reduce the size of plants by growth inhibition due to heat stress and shorten the overall growth period of crops, affecting the accumulative photosynthesis. These discussions are consistent with the result of shortened non-productive tillering stage in the AT + 3 °C treatment. This seems to have contributed to an earlier heading by about 4 and 2 days in 2016 and 2018, respectively. However, in both years, growth in the AT + 3 °C treatment was generally greater from the early growth stage to heading. This implies that growth inhibition by heat stress may not have significantly occurred in the AT + 3 °C treatment. The difference between height growth in the two temperature regimes was greater in 2016 (Figure 3A).

In contrast, SPAD and NDVI decreased more rapidly at AT. The SPAD values were significantly different (p < 0.01) between the two temperature regimes after 89 DAT and through the late growth period in 2016 (Figure 3B). The SPAD values in 2018 showed a significant difference (p < 0.05) between the two temperature regimes after 90 DAT and through the rest of the experiment, though the difference was slightly reduced at the last measurement (Figure 3D). In 2016 and 2018, the NDVI of AT declined rapidly after 83 DAT and 90 DAT, respectively, while those of the AT + 3 °C treatment declined gradually (Figure 3B,D). These results show that senescence is delayed by the residual assimilation products in the stem or leaves after heading since the higher temperatures at the moment of heading cause spikelet sterility [33]. This means that a higher chlorophyll content is retained for a relatively long period, thus maintaining the photosynthetic capacity. This seems to be related to a previous study’s finding that the distribution of dry matter to leaves and stems increased after the end of grain filling as the senescence of the panicle rapidly progressed at high temperatures [27].

3.2. Leaf Net Photosynthesis

A crop’s LUE can be used for spatiotemporal assessment and monitoring plant stress, as LUE_leaf decreases in stressed leaves compared to normal leaves [36,37]. There were no significant differences in LUE_leaf in both the tillering and booting stages in 2016 (p ≥ 0.05) (Figure 4A), but in the AT treatment, it was significantly lower at the grain filling stage (p < 0.01). In the overall patterns of LUE_leaf in 2016, under the AT regime, it showed relatively little difference from the tillering to the booting stage although in 2016 and 2018 it had a decreased and increased pattern, respectively. However, the LUE_leaf of AT decreased rapidly in the grain filling stage. Compared to AT, the LUEleaf in the AT + 3 °C regime slightly decreased in the grain filling stage as an extension of the decline from the tillering to the booting stage.

The LUE_leaf of the two temperature regimes at the tillering stage in 2018 showed no significant difference (p ≥ 0.05) (Figure 4C). However, there was a significant difference at the booting stage, when the LUE_leaf under AT increased above that of AT + 3 °C, which decreased. Unlike 2016, such a difference of LUE_leaf between the two temperature regimes in 2018 may be due to an extreme heatwave event in 2018. Way et al. [17] and Yamori et al. [38] reported that plants adapt to the environment even if the amount of photosynthesis decreases somewhat. Therefore, the relative decrease in photosynthetic capacity at AT + 3 °C at the booting stage of 2018 may have been the result of the small plant height difference between AT and AT + 3 °C. At the grain filling stage, it decreased in both air temperature regimes compared with that of the booting stage, but it decreased more rapidly in the AT treatment, leading to a significant difference (p < 0.05). Notably, the variation in the LUE_leaf values was quite large at this stage due to the difference of leaf chlorophyll content between the two temperature regimes (see Figure 3B,D).

The A_max at the booting stage was higher than that of the tillering stage in both years and decreased at the grain filling stage (Figure 4B,D). No difference between the two air temperature regimes was observed at the tillering stage, but A_max under AT was significantly higher than that at AT + 3 °C at the booting stage (p < 0.05). Specifically, it was about 9% and 4% lower at AT + 3 °C than at AT in 2016 and 2018, respectively. At the grain filling stage, A_max in the AT treatment decreased rapidly, whereas in the AT + 3 °C treatment, it decreased only slightly, resulting in a significantly higher A_max in the AT + 3 °C treatment (p < 0.01).

The changes in SPAD may have significantly related to photosynthetic capacity such as A_max. No significant differences in the SPAD values of the photosynthetic samples were found between the temperature treatments at the tillering and booting stages in both years (p ≥ 0.05). At the grain filling stage, the SPAD value at AT decreased compared to that of the previous stage, but at AT + 3 °C, it either decreased slightly or stayed at a similar level (p < 0.01) (see Figure 2B,D and Figure 3B,D). Thus, the A_max difference of two temperature regimes at the tillering and booting stages can be interpreted in a similar context to the change in SPAD: the effects of elevated air temperature were not expressed as physiological responses, but these effects seemed to be expressed cumulatively at the booting stage (Figure 4B,D).

Commonly, LUE_leaf and A_max at AT + 3 °C were higher than those at AT at the grain filling stage. Physiological heat stress and damage may cause lower grain weight and/or more spikelet sterility (see the values of grain filling rate and 1000-grain weight in

Table 1. Four yield components and total grain yield, above-ground dry matter (AGDM), and the harvest index (HI) under different air temperature regimes (TR) in 2016 and 2018. The values are means ± SE based on n = 3 samples. The ANOVA results at the bottom of the table show the significance levels for the main effects and their interaction.

YEAR	TR	No. of Panicles per m2	No. of Spikelets per Panicle	Grain Filling Rate (%)	1000-Grain Weight (g)	Grain Yield (g/m2)	AGDM (g/m2)	HI
2016	AT	291.21 ± 12.58	90.62 ± 2.70	90.37 ± 0.70	25.23 ± 0.17	596.38 ± 45.65	1388.27 ± 220.83	0.42 ± 0.01
	AT + 3 °C	277.18 ± 7.03	82.36 ± 5.28	0.84 ± 0.77	17.98 ± 0.71	3.48 ± 3.23	1310.72 ± 17.98	0.01 ± 0.00
2018	AT	269.28 ± 14.48	120.53 ± 3.06	87.68 ± 0.57	24.35 ± 0.12	688.50 ± 42.08	1680.29 ± 133.53	0.41 ± 0.01
	AT + 3 °C	267.30 ± 3.78	99.10 ± 7.87	7.31 ± 5.26	20.06 ± 0.37	45.13 ± 34.47	1625.22 ± 95.83	0.03 ± 0.02
ANOVA z	YEAR	*	**	*	ns	**	**	ns
	TR	ns	**	**	**	**	ns	**
	YEAR × TR	ns	*	**	**	ns	ns	ns

^z ANOVA results: **, p < 0.01; *, p < 0.05; ns, not significant (p ≥ 0.05).

). These will bring a loss of sink capacity in the grains, as evidenced by the relatively stable SPAD and NDVI values. Thus, the photosynthetic product stays in the leaves and stems instead of being reallocated to the grains, delaying leaf senescence.

3.3. Yield Responses to Elevated Air Temperature

Each yield component—the no. of panicles per m², no. of spikelets per panicle, grain filling rate, and 1000-grain weight—and the total grain yield, AGDM, and HI were investigated under the two temperature regimes (Table 1). Under the AT regime, all factors were significantly higher (p < 0.01) except for the number of panicles per m² and AGDM. These results seem to be due to the fact that the influence of water temperature at the tillering stage is larger than the influence of air temperature [28], although there is a relationship between air and water temperature. At AT + 3 °C, the number of panicles per m² was 5 and 0.7%, the number of spikelets per panicle was 9 and 18%, the grain filling rate was 99 and 92%, and the 1000-grain weight was 29 and 18% lower than at AT in 2016 and 2018, respectively. These results led to a reduction in grain yield of 99 and 93% and in HI of 98 and 93% in 2016 and 2018, respectively.

The most significant difference was identified in the grain filling rate due to heat stress-induced spikelet sterility. In 2018, when there was a strong heatwave, the grain filling rate dropped by 3% compared to 2016, even in AT. However, the heatwave in 2018 did not reduce grain yield in AT compared to 2016, which appears to be due to the increases in the number of spikelets per panicle and AGDM (See Table 1) with abundant solar radiation and near-full-scale sterility damage (see Figure 5C). On the other hand, photosynthetic capacity (i.e., LUE_leaf and A_max) at the AT + 3 °C treatment, which did not decrease even during the grain filling stage (see Figure 4B,D), was not of any assistance to grain yield against the damage of spikelet sterility. This is evidenced from the relatively lower values of 1000-grain weight at the AT + 3 °C treatment, compared to AT (Table 1).

3.4. Effects of Heat Stress

As a result of evaluating the effects of heat stress using the HDD index, the number of panicles per m² showed a negative relationship with the HDD (Figure 5A), while the number of spikelets per grain showed no statistically significant relationship with the HDD (p ≥ 0.05) (Figure 5B). The grain filling rate, grain yield, and HI decreased rapidly when the HDD exceeded 5 °C, and were close to 0 when it was above 10 °C (Figure 5C,E,F). This means that when the sum of temperatures greater than or equal to 35 °C per day that rice plants are exposed to from heading to maturity exceeds 5 °C, these factors rapidly decrease. The 1000-grain weight showed a significant negative relationship with the HDD index, with a coefficient of determination (R²) of 0.70 (Figure 5D). The number of panicles per m² and the number of spikelets per panicle, which are determined in the early growth stage, are dominated by water temperature rather than air temperature, while the grain filling rate and 1000-grain weight, determined in the late growth stage, are dominated by air temperature rather than water temperature [28]. Therefore, heat stress due to high temperatures during the early stage of growth may be mitigated by the lower water temperatures relative to air temperatures.

Changes in the growth and physiological responses of paddy rice were investigated under two air temperature regimes, AT and AT + 3 °C. During the vegetative growth period, the elevated air temperatures in the TGFC accelerated the expansion of leaves and stems. This is possible to result in increased light capture by the rice canopy, which can cause large amount of canopy photosynthesis. On the other hand, such rapid growth brought the heading date forward. The elevated temperature in the heading stage causes spikelet sterility, which greatly affects the grain filling rate and ultimately largely dominates grain yield.

4. Conclusions

The cumulative amount of photosynthesis can be reduced during the vegetative growth period. Further, the little reduced photosynthetic capacity found at the booting stages under high temperature stress may have a negative effect on grain yield. However, these effects do not seem to be critical on rice grain yield. Because of the sterility-derived delayed aging of leaves, the photosynthetic capacity in the grain filling stage was increased, but it could make up for the loss of grain yield due to spikelet sterility. In 2018, the heatwave event caused some abnormal rate of spikelet sterility even under AT in the TGFC. Given that spikelet sterility increases rapidly with an elevated temperature, the 2018 heatwave can be seen as a critical threshold condition that could damage rice grain yield. Therefore, future prediction studies on rice yield should consider the relationship between high temperature and spikelet sterility rates.