Punched-top chamber for moderately raising air temperature during the ripening period in rice

ABSTRACT Rice growth at an elevated air temperature (T air) during the ripening period is often evaluated using a semi-closed chamber (SCC). However, the water vapor pressure deficit (VPD) and CO2 concentration inside SCCs get lower than at ambient air plot, and these changes affect panicle temperature and photosynthesis. We developed a punched-top chamber (PTC), that is, an SCC with numerous pores on the top, and compared meteorological environments inside the two chambers and of ambient air plot. When solar radiation was >200 W m-2, ΔT air (SCC – Ambient) was 3.1°C–5.3°C, and ΔT air (PTC – Ambient) was 2.2°C–3.7°C. Excessively high T air > 38°C were more frequent inside the SCC than the PTC. The changes in VPD and CO2 concentration inside the PTC were less pronounced compared with those of the SCC, and thus PTC can be a better treatment for safely assessing the direct effect of elevated T air. Graphical abstract


Introduction
The increase in air temperature (T air ) due to the global climate change has various effects on crop production. The most commonly used method for raising T air during the ripening period in rice is a semi-closed chamber (SCC), which is a closed-top and open-sided plastic greenhouse. The SCC raises the daily maximum T air by 5.3°C-5.9°C and the daily mean T air by 2.5°C-3.0°C without a thermal source for 30 days after heading (Komaki et al., 2002); moreover, SCCs can be easily constructed at low cost (Chiba & Terao, 2014). However, SCCs excessively raise T air during the evaluation of the effects of high T air on grain filling and appearance quality. Komaki et al. (2002) reported that a 20-30 cm open-sided SCC raised the T air up to 40°C, which led to a significant increase in infertile grains as well as unmatured and dead grains. In this condition, an effect of the interaction between genotypes and T air on the percentage of undamaged grains was not observed, although this interaction is important for the selection of genotypes with heat tolerance. Ishizuki et al. (2013) developed an SCC with a length of the open side that was automatically controlled. This can avoid excessive heat stress on rice plants; however, the installation of the automatic control system requires more effort and is more costly than the simple SCC.
An open-top chamber (OTC), which was originally developed to evaluate the effect of high CO 2 and air pollution, also raises T air inside the chamber (Heagle et al., 1979). For grassland, the increases in T air observed inside OTCs differed according to plant species, chamber design, and environment (Leadley & Drake, 1993). The T air increase was proportional to the solar radiation (Whitehead et al., 1995), and the daytime increase ranged from 0°C to 4.3°C (Heagle et al., 1979;Messerli et al., 2015;Moya et al., 1997;Olszyk et al., 1980;Weinstock et al., 1982;Whitehead et al., 1995). In a flooded paddy field, Chiba and Matsumura (2006) reported that an OTC increased the daytime T air by as little as 0.5°C, which is lower than that detected for grassland. Thus, those authors developed new OTCs fitted with solar heated tunnels in subsequent studies (Chiba & Terao, 2014;Terao & Chiba, 2016); the new chambers increased the midday T air by about 1°C. However, the T air increase afforded by the OTC is smaller than that afforded by the SCC, and a simpler design would be preferable for the establishment of multiple chambers and for the evaluation of large amounts of plants, because the construction of heating tunnels also requires effort and additive materials.
The increases in T air inside both the SCC and the OTC are related to the fact that the cover of the chamber prevents gas exchanges between inside and outside the chamber. Therefore, the water vapor pressure deficit (VPD) and CO 2 concentration, in addition to T air , are different between inside and outside the chambers (Norby et al., 1997;Weinstock et al., 1982;Whitehead et al., 1995). To date, the manners in which VPD and CO 2 concentration inside the chambers are affected by the SCC design have been ignored during the evaluation of rice growth. In the other warming chamber, temperature gradient chamber, CO 2 concentration and VPD can be adjusted to match the outside environments by controlling the rates of air ventilation and CO 2 injection (Horie et al. 1995), but SCCs in a normal design have no adjustment function.
We developed a new chamber for raising T air during the ripening period of rice, namely the punched-top chamber (PTC). The PTC is an SCC with numerous pores on the top. The PTC with a design that is intermediate between the SCC and the OTC may afford an average T air , VPD, and CO 2 concentration. The objective of our study was to clarify the differences in VPD and CO 2 concentration as well as T air between the SCC, PTC and ambient air, to prove that the PTC is a useful alternative to the SCC for the evaluation of plant performance at high T air . As a guideline for the PTC to be an alternative, we aimed at least 2°C increase during midday inside PTCs compared with ambient air, a decreased frequency of excessive high T air (>38°C), and smaller changes in VPD and CO 2 concentration than inside SCCs. We also used the micrometeorology model, IM 2 PACT  to estimate effects of the microclimates inside the chambers on panicle temperatures (T pan ).

Plant cultivation
The experiments were conducted at the paddy field of the Central Region Agricultural Research Center, NARO, in Joetsu, Niigata, Japan (37°6΄N, 138°16΄E) in 2017 and 2018. We transplanted medium-sized seedlings of the Dontokoi cultivar to a paddy field on 16 May 2017 and 11 May 2018. The plant density was 22.2 plants m −2 (30 × 15 cm) at the rate of two plants per hill. All chemical fertilizers were applied as basal and were composed of 7 g N m −2 coated urea (LP100 that releases 80% of the nitrogen at a uniform rate until 100 days after application), 4 g m −2 P 2 O 5 , and 4 g m −2 K 2 O. Plants were grown under flooded conditions with 2-week middrainage from 30 days after transplanting, and were protected from diseases and insects using chemicals. The heading dates were August 4 and August 2, and plant maturity was achieved on September 18 in 2017 and September 15 in 2018, respectively.

Warming treatments
In a field, three treatment plots consisting of ambient air plot, SCC plot, and PTC plot were set with two replicates in 2017 and three replicates in 2018. Each plot of 1.8 m × 1.8 m was assigned in a completely randomized design in a field (>206 m −2 ) in two years. The distance between the plots was longer than 2.4 m. The SCC and PTC were installed from one week after heading dates to plant maturity. The SCC and PTC were developed by covering a rectangular metal frame with two transparent plastic films of the ceiling and side (0.1 mm thickness; Daiichi Vinyl, Fukui, Japan), with the frame being based on the design described by Chiba and Terao (2014). When it rained, the films of ceiling and side were separated under the weight of the accumulated water on top of the SCCs unlike PTCs, and the recorded data during these periods were excluded from the analysis. The sizes of the frames were a 1.8 × 1.8 m horizontal square with a height of 1.5 m for the SCC, and a -1.8 × 1.8 m square with a height of 1.3 m for the PTC (Figure 1). The 0.2 m lower height of the PTC was aimed at shortening the time of gas exchange between the inside and the outside of the chamber by decreasing the volume covered with the plastic films. The top of the PTC had approximately 162 pores (9 × 18) with a diameter of 3 cm and a spacing of 20 × 10 cm cut using a compass fitted with a razor. Preliminary tests suggested that (1) T air did not change when area-percentage of hole/top was the same even though the hole spacing and diameter were different (PTC-A vs. PTC-B, Table S1), (2) the halved area of the holes did not improve the increase in T air (PTC-A vs. PTC-C), and (3) a doubled area of the holes or increased open length of the side decreased the T air increase (PTC-A vs. PTC-D, PTC-E). In the present study, the open length of the side was set at 0.35 m for both PTCs and SCCs ( Figure 1) to enhance T air increase during the daytime, rather than the preliminary test side open length of 0.5 m (PTC-A). Light transmittance of the top film of SCCs measured by a light analyzer (LA-105; NK System, Osaka, Japan) was 94%, and the hole area of the top of the PTC was 3.5%. Thus, the holes of the PTC only increased light transmittance by 0.2% compared to SCC.

Collection of meteorological data
T air and relative humidity were measured at a central location of each ambient air plot, SCC plot, and PTC plot at a height of 1.0-1.1 m above the ground surface, which is the height of the panicles and flag leaves at heading. These measurements were done in two plots for each treatment in 2017 and 2018, although there was no replicate for ambient air plot in 2018, for the period from August 16 to September 14 in 2017 and from August 11 to September 13 in 2018.
T air and relative humidity were recorded by a sensor with logging function (Hygrochron; KN Laboratories, Inc., Osaka, Japan) in 2017, and by a sensor (HMP50 or HMP60; VAISALA, Helsinki, Finland) attached to a data logger (CR-10X; Campbell Scientific Inc., Utah, USA) in 2018. These sensors were set inside an aspirated radiation shield that moved air at 20 m 3 h −1 (Fan Aspirated Solar Radiation Shield 380-283; NovaLynx Corp, California, USA) in the two years. Exceptionally, one logger of two replicates for ambient air plot in 2017 was set inside another type of aspirated radiation shield, NIAES-09 (Fukuoka et al., 2011). T air and relative humidity were measured at 30 min intervals in 2017 and at 10 min intervals in 2018. VPD (hPa) was calculated by T air (°C) and relative humidity (RH, %) using the following equation (Murray, 1967). VPD ¼ 6:1078 � 10 7:5 Tair=ðTairþ237:3Þ � ð1 À RH=100Þ Solar radiation and wind speed at a height of 6 m were measured at 1 min intervals at a meteorological station located 300 m distant in a same experimental field. The wind speed measured at the meteorological station was significantly correlated with the wind speed measured at a height of 3 m in the tested paddy field (r = 0.649, P < 0.001), with the latter being 0.462-fold the former, on average ( Figure S1). The mean daily T air and mean daily solar radiation from heading to plant maturity were 24.6°C and 14.7 MJ m −2 d -1 in 2017, and 25.4°C and 14.2 MJ m −2 d -1 in 2018, respectively.

Calculation of panicle temperature
In order to estimate the effect of microclimate changes inside the plant canopy caused by the chamber application on T pan , the micrometeorology model, IM 2 PACT  was applied to the measured data at ambient air, PTC and SCC plots. IM 2 PACT is a physical model that solves for the heat balance inside the canopy obtaining the T pan as the solution, and the main input variables are T air , RH, solar radiation, and wind speed. The measured values of T air and RH inside the canopy of each treatment were directly given as input values. For solar radiation, the value of the meteorological station was given for the ambient air plot, and the values of 94.2% and 94% of it were given for the PTC and SCC plots, respectively, according to the light transmission characteristics of the chamber film. The input value of wind speed above the canopy for ambient air plot was given by the value at 6 m height of the meteorological station multiplied by the above ratio of 0.462. The wind speeds for PTC and SCC plots were set assuming that the chamber application would reduce the wind level to 1/10 of the level in the ambient air plot. Since there is no actual measurement of wind speed in the chamber, the reduction to 1/10 was set as a reference, but the actual wind speed may be lower than that because there is almost no wind in the chamber. For the panicle transpiration conductance and the bulk canopy transpiration conductance, the values of Akitakomachi cultivar  were applied as typical values for Japanese cultivars because of no actual measurements.

Measurements of CO 2 concentration inside the chambers
The CO 2 concentrations near panicles and flag leaves were measured in the center of the ambient air plot, SCC plot, and PTC plot using three replicates in 2018. The measurements were made in the daytime from August 24 to 30 in 2018, and the clock time during the measurement was different by day in order to vary solar radiation and wind speed. The air was aspirated at a height of 1.1 m by a flow control pump (minipump MP-Σ300, SIBATA scientific technology LTD, Tokyo, Japan) at the rate of 1.0 L min −1 through a nylon tube, and the air for each plot was manually switched using a three-way valve. Next to the pump, the air was introduced into a 200 mL Erlenmeyer flask, to lessen the fluctuation of CO 2 concentration by stirring the air, and passed through a 1-μm air filter (Merk Millipore, Mass., USA) and a membrane gas dryer (SWC-M04-70/ IP; AGC Engineering CO. LTD., Chiba, Japan). The CO 2 concentration was measured by an infrared gas analyzer (LI-820; LI-COR, USA) at 5 s intervals, and the concentration at each plot at a given time was determined as a mean for three consecutive recorded values at a stable state. Because the volumes of the SCC and PTC covered with plastic film were larger than 3078 L, the effect of gas absorption at the rate of 1.0 L min −1 on the measured values of CO 2 concentration was negligible.

Statistical analysis
Analysis of variance (ANOVA) and Tukey's post hoc (Tukey-HSD) tests for multiple comparisons among the treatments were performed on the values for T air , VPD, T pan , and CO 2 concentration. ANOVAs were performed using a mixed model, with treatment and year as dependent variables, and measurement time as a random effect for T air , VPD and T pan . For CO 2 concentration, the model had treatment as a dependent variable and measurement time as the random effect. Statistical analyses were carried out using JMP 14.3.0 (SAS Institute Inc., Cary, NC, USA) statistical software.

Results
In 2017, the solar radiation during the warming treatments from August 10 to September 16 was comparative to the mean for the 10 years from 2007 to 2016, while in 2018, solar radiation was below the mean till noon (Figure 2). The solar radiation greater than 200 W m -2 is expected from 8:00 to 16:00 in the 10 years, and from 10:00 to 14:00 for the 25th percentile. When solar radiation was larger than 200 W m -2 , there was a significant difference in T air (p < 0.05), with SCC, PTC, and ambient air plot being higher in that order (Table 1). The ΔT air (SCC -Ambient) and ΔT air (PTC -Ambient) were enhanced under a greater solar radiation and smaller wind speed. When solar radiation was 0 W m − 2 at nighttime, the T air inside the two chambers was 0.6°C-1.4°C lower than that detected in the ambient air plot. Similarly to T air , the calculated T pan values inside the SCC and PTC were significantly higher than the ambient air plot when solar radiation >0 W m -2 (p < 0.05, Table 2). The ΔT pan (SCC -Ambient) and ΔT pan (PTC -Ambient) were 2.5°C-3.6°C greater than the respective ΔT air values when solar radiation >200 W m -2 . Figure 3 shows the frequency distribution of T air inside the SCC and PTC when T air in ambient air plot was higher than 30°C and solar radiation ≥200 W m -2 . Frequency of high T air was higher in the SCC than the PTC; frequencies of >38°C were 33.3% in 2017 and 7.8% in 2018 in the SCC, and those of PTC were 3.9% and 1.0%, respectively. The daily mean T air on fine days with a daily solar radiation >20 MJ m −2 d -1 was 1.8°C higher in the SCC and 1.0°C higher in the PTC compared with the ambient air plot (Table 3). An increase in daily mean T air was not observed on days with a daily solar radiation <15 MJ m −2 d -1 .
The changes in VPD inside the SCC and PTC varied with environmental conditions (Table 4). The VPDs inside the SCC were significantly higher than the ambient air plot at wind speed <2 m s −1 and solar radiation >200 W m −2 (p < 0.05). In the other conditions, the VPDs inside the SCC were significantly lower than those of the ambient air plot (p < 0.05). In both situations, the PTC exhibited a similar or smaller difference in VPD vs. ambient air plot, compared with the SCC. The difference in T air and VPD between the two replicates of measurement tended to get larger when solar radiation >0 W m -2 compared to solar radiation = 0 W m -2 (Table S2).
The CO 2 concentration inside the warming chambers was significantly lower than that detected in the ambient air plot (p < 0.05, Table 5). The decreases in CO 2 concentration inside the warming chambers were greater at wind speed <2 m s -1 . At this condition, CO 2 concentration inside the SCC was significantly lower than that inside the PTC, and the CO 2 concentrations Table 1. Air temperature near panicles in the ambient air plot (Ambient) and inside the warming chambers at different solar radiations (S, W m −2 ) and wind speeds in 2017 and 2018.  inside the SCC and the PTC were 86 μmol mol -1 (21%) and 46 μmol mol -1 (11%) lower than that in ambient air plot, respectively (p < 0.05).

Discussion
The SCC exhibited not only increases in T air (Table 1) but also changes in VPD (Table 4) and decreases in CO 2 concentration (Table 5), which indicates that the environmental factors including VPD and CO 2 concentration other than T air may affect grain filling of the plants inside the SCC. A decrease in CO 2 concentration around the plant canopy decreases the photosynthesis rate (Farquhar & Sharkey, 1982), and a reduction in the carbon assimilate supply during grain filling increases the percentage of chalky grains (Tsukaguchi & Iida, 2008). The dewdrops present at the top when VPD is small and ventilation is low reduces light transmittance (Harazono et al., 1997), and dewdrops were often observed on the cover of an SCC (data not shown). The microclimate inside the PTC was intermediate between those of the SCC and ambient air plot; the reduction in CO 2 concentration inside the PTC was almost half of that in the SCC at low wind speed <2 m s -1 (Table 5), and the changes in VPD inside the PTC were similar to or smaller than those of the SCC (Table 4). Unlike SCCs, almost no dewdrops were observed at the top of PTCs, probably due to ventilation through pores. The ΔT air (PTC -Ambient) was 2.2°C-3.7°C when the solar radiation was > 200 W m -2 (Table 1), which was larger than 1.0°C, reported for the improved OTC fitted with solar heated tunnels (Chiba & Terao, 2014;Terao & Chiba, 2016). There was less opportunity for T air inside the PTC to reach excessively high values >38°C (1.0-3.9%) than in the SCC (7.8-33.3%), when the outside T air was >30°C (Figure 3). The excessively high T air of >38°C around flowering time increases the percentage of infertile grains (Matsui, 2009), which complicates the evaluation of grain filling and apparent quality (Komaki et al., 2002). In addition to the decreased frequency of high T air , the PTC achieved the target values of T air and had less changes in VPD and CO 2 concentration (Tables 4  and 5), and thus the PTC was determined to be a better method than the SCC for safely assessing the direct effect of high T air .
Plant body temperature, such as T pan , is greatly affected by environmental factors other than T air , especially by humidity and wind speed through the effect of evaporative cooling of the plant body . For both SCC and PTC, the chamber application  The rain fall in the listed days was < 5 mm. The daily temperature of the days with Foehn and Typhoon winds was excluded from the calculation of average temperature. Here, Foehn wind is defined as a 10 min average wind stronger than 3 m s −1 for at least 2 h. The temperatures of fine days from 17 to 21 August 2018 had failed to be record. Values within the parenthesis represent the difference from Ambient.
blocks the diffusion of water vapor from plant transpiration and water surface evaporation to the air above the chamber, so the absolute humidity inside the chambers always increased compared to ambient air plot at all solar radiation and wind speed conditions (data not shown). At solar radiation ≥200 W m −2 and wind speed <2 m s −1 , the increase in T air inside both the SCC and PTC relative to ambient air plot was large (especially inside the SCC, with an extreme increase of 5.3°C; Table 1), where VPD apparently increased slightly (Table 4) due to the increase in saturated water vapor pressure. However, at other solar radiation and wind speed conditions, VPD decreased (Table 4). The applying of these T air and humidity changes inside the SCC and PTC to IM 2 PACT with the assumption of the wind speed to be reduced to 1/10 of that of ambient air plot showed that the increase in T pan by the chamber application was greater than that of T air for both the SCC and PTC in terms of heat balance (Table 2). This suggests that the chamber application increases T pan , more than the increase in T air , which is applicable for investigating physiological responses to T air increase in the ripening stage of rice canopy. We note, however, that the wind speed in both SCC and PTC was set to be 1/10 of the ambient air plot in this simulation, so the increase in T pan was calculated to be similar between SCC and PTC ( Table 2), but the actual wind speed could be lower, especially in SCC since there was almost no wind, and the increase in T pan could be even greater in SCC than this simulation. In our study, while the T air inside both the SCC and PTC was higher than that in the ambient air plot when the solar radiation was >0 W m -2 in the daytime, the T air was lower inside the chambers when solar radiation was 0 W m -2 in the nighttime (Table 1); similar differences were observed for T pan (Table 2). T air changes in a day occur near the water surface and plants first, and then the T air change is transmitted to the above air (Geiger & Stewart, 1950). When the chambers are applied into the plant canopy, the heat balance in nighttime is such that the covers of the chamber themselves, in addition to the plant canopy, act as a cold source against the warmer above air. The observed T air difference at night, where the temperature inside the chambers was lower than that in the ambient air plot, is because the SCC and PTC plots are surrounded by more cold sources (the cooled cover of the chambers as well as the plant canopy) than the ambient air plot. The phenomenon of lower T air than the outside at night in greenhouse or chamber without heating system has been reported in the previous studies (Chiba & Terao, 2015;Harazono et al., 1997;Terao & Chiba, 2016). These results indicate that the SCC and PTC in this study are not suitable for experiments focusing on high night temperatures. Although ΔT air (SCC -Ambient) and ΔT air (PTC -Ambient) were proportional to solar radiation in Table 4. Water vapor pressure deficit near panicles for the ambient air plot (Ambient) and inside the warming chambers at different solar radiations (S, W m -2 ) and wind speeds in 2017 and 2018.  daytime, the negative ΔT air (warming chambers -Ambient) was observed when wind speed >4 m s -1 in 2018 ( Figure S2), which would have been due to the prevention of warm foehn wind induction to the chambers. Thus, the effects of warming chambers on daily mean T air can vary depending on the characteristics of regional climate. The T air decrease in the night had a negative effect on the daily T air rise, but in total, daily T air inside the PTCs was 1.0°C higher than that in ambient air plots when daily solar radiation >20 MJ m −2 d -1 (Table 3). When daily solar radiation <15 MJ m −2 d -1 , no significant increase in daily T air was observed. These results can be attributed to the T air increase during the daytime being proportional to the solar radiation inside PTCs as well as SCCs ( Figure S2), as previously reported (Terao & Chiba, 2016;Whitehead et al., 1995). Thus, the PTC and SCC are effective in increasing the T air in regions with a strong solar radiation. The frequency of days with fine weather varies according to regions; fine weather with daily solar radiation >20 MJ m −2 d -1 is expected for an average of 46% days in August in Japan, that is, the frequency of fine days >20 MJ m −2 d -1 in August for the 10 years (from 2007 to 2016) was 45% for the experimental site, Joetsu (37°6΄N), 41% for the Northern site, Daisen, Akita prefecture (39°5΄N), and 55% for the southern site, Fukuyama, Hiroshima prefecture (34°5΄N). The frequency of daily solar radiation <15 MJ m −2 d -1 for the same period was 25%−38% for the three sites.
We demonstrated that the PTC, which had numerous pores on the top of the SCC, exhibited at least 2°C increase in T air when solar radiation was greater than 200 W m -2 , with lesser frequency of excessive high T air , and less pronounced changes in VPD and CO 2 concentration compared with the SCC. The changes in microclimates inside the PTC are affected by the air flow through the lower open spaces at the side, as well as from pores on the top, as seen from the fact that relationship between T air inside SCC and solar radiation is affected by wind speed ( Figure S2). Thus, it is important to note that when installing PTCs, the targeted increase in T air should be adjusted by controlling the lower open spaces and height and volume of the PTC according to the regional climate and structure of the plant canopy.