Source–sink dynamics and proteomic reprogramming under elevated night temperature and their impact on rice yield and grain quality


  • Wanju Shi,

    1. College of Agronomy, Hunan Agricultural University, Changsha, Hunan, China
    2. Crop and Environmental Sciences Division, International Rice Research Institute (IRRI), Metro Manila, Philippines
    Search for more papers by this author
  • Raveendran Muthurajan,

    1. Centre for Plant Molecular Biology, Tamil Nadu Agricultural University, Coimbatore, Tamil Nadu, India
    Search for more papers by this author
  • Hifzur Rahman,

    1. Centre for Plant Molecular Biology, Tamil Nadu Agricultural University, Coimbatore, Tamil Nadu, India
    Search for more papers by this author
  • Jagadeesh Selvam,

    1. Centre for Plant Molecular Biology, Tamil Nadu Agricultural University, Coimbatore, Tamil Nadu, India
    Search for more papers by this author
  • Shaobing Peng,

    1. MOA Key Laboratory of Crop Ecophysiology and Farming System in the Middle Reaches of the Yangtze River, College of Plant Science and Technology, Huazhong Agricultural University, Wuhan, Hubei, China
    Search for more papers by this author
  • Yinbin Zou,

    1. College of Agronomy, Hunan Agricultural University, Changsha, Hunan, China
    Search for more papers by this author
  • Krishna S. V. Jagadish

    Corresponding author
    1. Crop and Environmental Sciences Division, International Rice Research Institute (IRRI), Metro Manila, Philippines
    • College of Agronomy, Hunan Agricultural University, Changsha, Hunan, China
    Search for more papers by this author


This article is corrected by:

  1. Errata: Corrigendum Volume 203, Issue 2, 704, Article first published online: 13 May 2014

Author for correspondence:

Krishna S. V. Jagadish

Tel: +63 9175648526



  • High night temperatures (HNTs) can reduce significantly the global rice (Oryza sativa) yield and quality. A systematic analysis of HNT response at the physiological and molecular levels was performed under field conditions.
  • Contrasting rice accessions, N22 (highly tolerant) and Gharib (susceptible), were evaluated at 22°C (control) and 28°C (HNT). Nitrogen (N) and nonstructural carbohydrate (NSC) translocation from different plant tissues into grains at key developmental stages, and their contribution to yield, grain-filling dynamics and quality aspects, were evaluated. Proteomic profiling of flag leaf and spikelets at 100% flowering and 12 d after flowering was conducted, and their reprogramming patterns were explored.
  • Grain yield reduction in susceptible Gharib was traced back to the significant reduction in N and NSC translocation after flowering, resulting in reduced maximum and mean grain-filling rate, grain weight and grain quality. A combined increase in heat shock proteins (HSPs), Ca signaling proteins and efficient protein modification and repair mechanisms (particularly at the early grain-filling stage) enhanced N22 tolerance for HNT.
  • The increased rate of grain filling and efficient proteomic protection, fueled by better assimilate translocation, overcome HNT tolerance in rice. Temporal and spatial proteome programming alters dynamically between key developmental stages and guides future transgenic and molecular analysis targeted towards crop improvement.


On a global scale (Vose et al., 2005; IPCC, 2007) and at the farm level (Peng et al., 2004), minimum night temperatures are increasing at a much faster pace than maximum day temperatures, and this trend is projected to continue into the future (Christensen et al., 2007). Controlled environment studies (Cheng et al., 2009; Mohammed & Tarpley, 2009a,b; Kanno & Makino, 2010), as well as field experiments (Peng et al., 2004; Nagarajan et al., 2010), have recorded a significant negative impact of higher minimum night temperature on rice yield. Hence, efforts must be intensified to address this emerging phenomenon in synchrony with the progress being achieved in breeding for high day temperature tolerance in rice mega varieties (Jagadish et al., 2010; Ye et al., 2012) to induce diurnal temperature tolerance in rice. To achieve this target, a diverse set of entries must be tested for their response to high night temperatures (HNTs), which is a prerequisite to the identification of contrasting entries in order to better understand and explore the physiological and molecular mechanisms that induce tolerance.

The yield penalty under HNT has been attributed to a reduction in the number of panicles per square meter (Peng et al., 2004), final grain weight (Morita et al., 2005; Kanno & Makino, 2010) and spikelet fertility (Cheng et al., 2009; Mohammed & Tarpley, 2009a, 2010), which are partly explained by increased respiration rates, membrane leakage (Mohammed & Tarpley, 2009b) and reduced pollen germination (Mohammed & Tarpley, 2009a). However, the majority of the conclusions drawn above are based on individual genotype performance – for example, IR72 (Cheng et al., 2009), Cocodrie (Mohammed & Tarpley, 2009a,b, 2010) and Akita-63 (Kanno & Makino, 2010); almost all of these studies were conducted under controlled environments. Therefore, there is a significant gap in the identification of contrasting rice genotypes and their physiological and molecular responses on exposure to HNTs under realistic field conditions.

Temperature at night has been speculated to have an impact on the flowering dynamics on the following morning (Kobayashi et al., 2010), but it has not been studied systematically. Photoassimilates generated either during grain filling (post-anthesis) or redistributed from the reserve pool of the vegetative tissues (pre-anthesis) determine successful grain filling in rice (Yang & Zhang, 2006). Limited information is available on the effect of HNT on dry matter production, carbohydrate (sugars and starch) and nitrogen (N) partitioning, and grain filling, which are critical determinants of final grain yield. Final grain weight is determined by the rate and duration of grain filling in rice. However, the magnitude of change with HNT on the rate and duration of grain filling in contrasting rice genotypes has not been estimated. The above-mentioned sequence of yield-influencing processes could have a major influence on grain quality, which is increasingly becoming an essential determinant of the market price, and thus warrants detailed investigation.

To capture the impact of extreme temperatures and other abiotic stresses at the molecular level in rice, the proteomic (two-dimensional gel electrophoresis) approach has been effectively employed (Cui et al., 2005; Jagadish et al., 2010, 2011). However, in the majority of the studies, either vegetative (Salekdeh et al., 2002; Yan et al., 2005) or reproductive (Imin et al., 2004; Liu & Bennett, 2011) tissues, and generally at a single time point, have been used to study proteome changes. Yan et al. (2005), using salt stress-affected rice seedling roots, and Kerim et al. (2003), using anthers at different developmental stages, applied the two-dimensional proteomic approach and demonstrated the proteome dynamics at different time points. To our knowledge, no reports have addressed the proteome changes with HNT using both vegetative and reproductive tissues at economically relevant time points, such as flowering and early grain filling (EGF).

Unlike all the above-mentioned studies, our trial was carried out using temperature-controlled chambers under field conditions. Preliminary wide genetic diversity screening for HNT among 36 different rice accessions using the above-mentioned field chambers formed the basis of this experiment, from which the most contrasting entries were selected for physiological and molecular characterization. Both field and laboratory analyses were undertaken as follows: to estimate the impact of HNT on grain yield and yield components between two contrasting rice genotypes under field conditions; to quantify N, nonstructural carbohydrate (NSC) and biomass partitioning at key developmental stages in response to HNT; to determine the impact of HNT on flowering pattern, rate and duration of grain filling along different sections of the panicle, and grain quality; and to unravel the temporal reprogramming of the flag leaf and spikelet proteome exposed to HNT at flowering and EGF, and to establish their relevance to physiological responses.

Materials and Methods

Field experiment and laboratory analyses were conducted in 2011 at the International Rice Research Institute (IRRI), Los Baños (14°11′N, 121°15′E, 21 m asl), the Philippines.

Crop husbandry

Two contrasting rice (Oryza sativa L.) genotypes, N22 with high temperature tolerance and Gharib with high temperature sensitivity, were chosen for this study based on data obtained from previous genotypic diversity analyses comprising 36 rice accessions, exposed to 26°C (HNT) and 22°C (control) (Zhang et al., 2012). Seed dormancy was broken by exposure to 50°C for 3 d, followed by pre-germination and sowing in seeding trays. Fourteen-day-old seedlings were transplanted on 22 June 2011 at a spacing of 0.2 m × 0.2 m with four seedlings per hill. Phosphorus (15 kg P ha−1 as single superphosphate), potassium (20 kg K ha−1 as KCl) and zinc (2.5 kg Zn ha−1 as zinc sulfate heptahydrate) were applied and incorporated into all plots 1 d before transplanting. N fertilizer in the form of urea was applied in four equal splits (30 kg ha−1 as basal, 20 kg ha−1 at mid-tillering, 20 kg ha−1 at panicle initiation and 30 kg ha−1 just before heading). Manual weeding was employed to maintain weed-free plots. Whorl maggots (Hydrellia philippina Ferino) during the early vegetative stage and yellow stem borers (Scirpophaga incertulas) at the flowering stage were effectively managed by chemical spraying.

HNT chambers and treatment

Twelve temperature-controlled chambers were specifically designed under field conditions to study the impact of HNT. Each chamber (6 m × 3 m × 2 m in length, width and height, respectively) were fixed with a 2.8-m interval to ensure adequate ventilation and to avoid mutual shading. The framework of the chambers consisted of a series of shed-type pipes (Supporting Information Fig. S1). Each chamber was equipped with an air conditioner (CW-1805V; Matsushita Electric Philippines Corp., Taytay, Rizal, the Philippines) capable of maintaining constant temperatures. There were two inlet and two outlet fans installed in the front frame and back frame, respectively, to minimize the differences in relative humidity (RH) and CO2 concentration within the chamber compared with the ambient by constant but mild air exchange. Stand-alone sensors were placed above the canopy (at 100 cm above the soil) in each of the chambers to measure temperature and RH once every minute and averaged over 30-min intervals, with all the sensors connected to data loggers (HOBO; Onset Computer Corp., Bourne, MA, USA). During the daytime (06:00–18:00 h), the chambers were open, exposing the plants to natural conditions. At night (18:00–06:00 h), the chambers were closed manually and the air conditioners were programmed to automatically impose control (22°C) and stress (28°C) treatments. Six replicate chambers each were used to impose the temperature treatments. Nearly 5 cm of standing water was maintained throughout the experiment to ensure a leak-proof covering of the chambers for the whole night. Temperature treatments started from the panicle initiation stage, c. 31 d after transplanting, and continued up to physiological maturity.


Growth analysis

At key developmental stages after the imposition of heat stress, 12 hills from each replicate chamber and variety were taken to determine biomass accumulation. Plants were separated into leaves, stem + sheath at panicle initiation and, additionally, panicles at flowering and 15 d after flowering (DAF). All plant samples were oven dried at 70°C for 5 d until a constant weight was recorded.

Grain yield and yield components

At physiological maturity, a central 2-m2 area (50 hills) in each chamber was sampled for grain yield analysis and the data were adjusted to the standard moisture content of 0.14 g H2O g−1.

Twelve hills (0.5 m2) were taken from each plot to determine the above-ground total dry weight and yield components. The panicle number was counted in each hill to determine the panicle number per square meter. Plants were separated into straw and panicles. Panicles were hand-threshed and the filled spikelets were separated from unfilled spikelets by submerging them in tap water. Three subsamples of 30 g of filled spikelets and 2 g of empty spikelets and all of the half-filled spikelets were taken to count the number of spikelets (Peng et al., 2010). The dry weights of straw, rachis and filled and half-filled spikelets were determined after oven drying at 70°C to constant weight. The above-ground total dry weight was the combined dry matter of straw, rachis, and filled, half-filled and empty spikelets. The number of spikelets per panicle, number of spikelets per square meter, seed set% (100 × (number of filled spikelets + number of half-filled spikelets)/total number of spikelets) and 1000-grain weight were calculated.

Flowering pattern

The main tillers of four plants, from each replicated chamber for N22 and Gharib exposed to HNT and control, were tagged to record the daily flowering pattern starting from the day of anthesis (at least one spikelet with protruding anthers) and continuing for three consecutive days. On each day, the number of spikelets undergoing anthesis was recorded every 30 min, starting from 08:30 h to 14:00 h, following cumulative counts to avoid manual interference.

NSC and N content

Plant samples were taken at 05:00 h just before the chambers were opened at panicle initiation, flowering, 15 DAF and at physiological maturity for NSC and N content estimation following Yoshida et al. (1976) and Bremner & Mulvaney (1982), respectively. To avoid confounding factors across early and late tillers in a hill, four main tillers from each hill (each hill had four seedlings) were selected with four replicates for each chamber. Tillers were separated into leaves, stems + sheaths and panicles, and immediately treated with a heat burst in the microwave for 1 min (Pelletier et al., 2010), and then dried at 70°C for 48 h. The samples were then ground and taken for NSC and N estimation (0.1 g each).

Rate and duration of grain filling

About 60 panicles on the main tillers that headed on the same day were tagged for each treatment. Starting at 100% anthesis, five tagged panicles were sampled randomly; the process was repeated once every 4 d until maturity. The panicles were divided into three equal parts (top, middle and bottom) based on panicle length. All grains, except the unfertilized spikelets, were weighed after oven drying at 70°C for 72 h to obtain constant dry weight. The grain-filling rate of the top, middle and bottom sections of the panicle were fitted using the logistic equation y = K/(1 + e− bx) (Brdar et al., 2008; Huang & Zou, 2009), where y represents the observed grain dry weight, x is the time after flowering, K is the estimate of the final grain weight, and a and b are parameters of the equation with only mathematical meaning, which were used to calculate the secondary parameters of grain-filling processes as mentioned below. R2, which is the correlation coefficient of the equation, was also estimated. The initial grain-filling rate, GR0 = Kbea/(1 + ea)2, maximum grain-filling rate, GRmax = Kb/4, mean grain-filling rate, GRmean = Kb/(a  loge(100/99 − 1)), time to reach the maximum grain-filling rate, Tmax = a/b, and the active grain-filling duration (D), were estimated with y at 95% of K and solving for X using the following equation: D = [X = (2.944 + a)/b].

Grain quality parameters

Eight replicate samples of seeds from each treatment and genotype were separated and analyzed for amylose content, protein content, chalkiness (0–10%, 1025%, 2550%, 5075% and > 75%), grain length and width at the Grain Quality and Nutrition Center, IRRI, Philippines. The physical characteristics of the grain were measured using the 1625 Grain Inspector (DK-3400 Hillerod, Foss, Denmark). To measure amylose, polished grains were ground to pass through a 0.5-mm sieve in a cyclone mill (Udy Cyclone Sample Mill 3010-030; Fort Collins, CO, USA). Amylose concentration was measured as described previously (Juliano, 1971). In addition, 125 g of seeds were used to estimate the brown rice (after removing the hull), head rice yield (percentage of grains with ≥ 3/4 the size of the original grain size) and percentage milled rice (head rice yield + broken grains) (Cooper et al., 2006).

Proteomic analysis of HNT response in flag leaf and spikelets

Two-dimensional polyacrylamide gel electrophoresis (PAGE)

Total soluble proteins were extracted from flag leaves and spikelets (after careful exclusion of the rachis and pedicle) from three replicates collected over two time points (100% flowering and 12 DAF) from both control and HNT-treated plants by the trichloroacetic acid precipitation method (Salekdeh et al., 2002). For spikelets collected at 12 DAF, which were at the early grain-filling stage, 2 M thiourea was added to the lysis buffer in addition to urea to solubilize the proteins thoroughly. All further processes, including protein quantification, isoelectric focusing and sodium dodecylsulfate (SDS)-PAGE, were carried out as described by Jagadish et al. (2010); 150 μg of proteins were loaded/rehydrated in pH 4–7 (length, 17 cm) immobiline pH gradient (IPG) strips and separated during the first dimension by isoelectric focusing (GE Healthcare, Wisconsin, USA). Proteins were further separated on the basis of their molecular weight on 12% SDS-PAGE gel.

Image acquisition, data analysis and protein identification

Silver-stained gels were scanned using an ImageScanner-III (GE Healthcare, Wisconsin, USA) with a resolution of 600 pixels and 16 bits per inch. Image visualization, spot detection and protein quantification were carried out using the Image Master 2D Platinum Version 6.0 (GE Healthcare, Wisconsin, USA). After automated detection and matching, manual editing for individual spots was carried out. The percentage volume of each spot was estimated and the abundance ratio (% volume of spot under stress/% volume spot under control; Yan et al., 2005; Jagadish et al., 2010, 2011) was calculated. Internal molecular markers were used to determine the experimental pI (isoelectric point) and molecular weight for the proteins of interest. The percentage volume from three replicates of HNT gels was used to check for significant variation in expression compared with data obtained from the same number of gels for the control. Protein spots changing by > 1.5-fold or more, and with statistical significance at 5% (P < 0.05) between control and temperature-treated tissue, were used for matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF) analysis. Peptide sequences obtained from MALDI-TOF MS were searched in MASCOT ( and Profound ( databases to identify proteins. The searches showing the highest MASCOT score with maximum sequence coverage were taken into account. Later, the protein sequences obtained from the database were searched in the TIGR database using the protein BLAST tool (, and their respective functions in rice were obtained.

Statistical analysis

Growth parameters, flowering pattern, yield and yield components, grain quality parameters, and NSC and N content were analyzed using Genstat 14th edition (Rothamsted Experimental Station, Harpenden, UK). The grain-filling rate and other associated parameters were estimated by nonlinear equation fitting using Microsoft Excel solver. Protein abundance (% volume) values across treatments and replications obtained from Image Master 2D Platinum software were analyzed as a completely randomized design using Genstat 14th edition.


Temperature and RH

The temperatures in the chambers were close to the set targets of 22°C (actual, 22.1°C; SD = ± 0.67°C) and 28°C (27.7°C; ± 0.81°C), and RH in the 22°C chambers was 97.2% (± 2.18%), whereas that in the 28°C chambers was 88.6% (± 3.11%). The temperature and RH during the day were similar to those in outside natural conditions – 28.1°C (± 2.57°C) and 87.3% (± 7.79%) in the 22°C chambers, and 28.6°C (± 2.19°C) and 85.9% (± 7.12%) in the 28°C chambers.

Flowering pattern, yield and yield components

The flowering pattern in both the tested entries showed no significant variation with HNT (P > 0.05) across three consecutive flowering days (Fig. S2). However, the two entries behaved differently with regard to the number of spikelets opening at peak anthesis: N22 recorded a smaller number of open flowers, whereas Gharib had more open flowers with HNT compared with the control, but these were not significantly different (P > 0.05). Yield and yield components and total dry weight at flowering and maturity were significantly different among the two genotypes (P < 0.05 to P < 0.001; Table 1). HNT induced significant differences with regard to grain yield, spikelets per square meter, seed set, 1000-grain weight, total dry weight and plant height at maturity (P < 0.05 to P < 0.01). Temperature and genotype interaction was significant only with grain yield (P < 0.05), seed set and 1000-grain weight (P < 0.001). Specifically, HNT reduced grain yield, 1000-grain weight and total dry weight at maturity by 21.8%, 7.9% and 13.5%, respectively, in the sensitive Gharib, whereas N22 was not affected. However, HNT decreased the number of spikelets per square meter by 14.6% and increased seed set by 7.6% in N22; these traits were not affected in Gharib.

Table 1. Growth, yield and yield components of rice (Oryza sativa) accessions N22 and Gharib exposed to control (22°C) and high (28°C) night-time temperatures from panicle initiation to physiological maturity
GenotypeTemperature (°C)Grain yield (g m−2)Panicles m−2Spikelets panicle−1Spikelets m−2 (*1000)Seed set (%)1000-grain weight (g)TDW at flowering (g m−2)TDW at maturity (g m−2)Plant height (cm)
  1. *, **, ***, Significance at 5%, 1% and 0.1%, respectively. LSD, least significant difference; ns, nonsignificant; TDW, total dry weight. Mean value ± SE.

N2222287.4 ± 3.7375 ± 859 ± 221.9 ± 0.681.1 ± 0.616.8 ± 0.1679.3 ± 35.9914.1 ± 15.3147.4 ± 1.9
28275.8 ± 9.0371 ± 350 ± 318.7 ± 1.087.3 ± 1.116.9 ± 0.3660.5 ± 23.8847.7 ± 22.7157.0 ± 1.8
Gharib22260.4 ± 7.8323 ± 846 ± 115.8 ± 0.679.7 ± 0.821.5 ± 0.1506.7 ± 11.1789.8 ± 55.6122.9 ± 1.6
28203.6 ± 11.5315 ± 948 ± 115.2 ± 0.577.7 ± 0.819.8 ± 0.1505.2 ± 13.0683.1 ± 21.6125.0 ± 2.3
LSD 29.714.
Genotype (G)  *** *** * *** *** *** *** ** ***
Temperature (T)  ** nsns * * ** ns * *
G × T  * ns * ns *** *** nsnsns

Grain quality parameters

All grain quality traits, including the brown, milled and head rice yields, were influenced significantly by genotype (P < 0.001; Table 2). The effects of temperature and the interaction between genotype and temperature were significantly different in all the traits (P < 0.05 to P < 0.001), except for head rice yield and amylose content (Table 2). An inherently low head rice recovery was observed in Gharib. HNT reduced the brown and milled rice yields by 2.0% and 4.0%, respectively, in Gharib compared with control; in N22, these two traits were unaffected. Similarly, grain width and protein content followed the same trends, with a reduction of 2.7% and 4.8%, respectively, in Gharib. N22 recorded a significant increase in grain length, which was the only measured trait not affected by HNT in Gharib. Although the chalk content in grains was not affected with different categories up to 50%, Gharib under HNT recorded a 56.6% decrease in chalk content with the 50–75% category, but showed a 36.4% increase in chalkiness with the > 75% category (Table 2).

Table 2. Effects of high night-time temperature on grain quality and brown, milled and head rice (Oryza sativa) yields of accessions N22 and Gharib
GenotypeTemperature (ºC)Brown rice (%)Milled rice (%)Head rice (%)Grain length (mm)Grain width (mm)Amylose content (%)Protein content (%)Chalkiness 50–75%Chalkiness > 75%
  1. *, **, ***, Significance at 5%, 1%, and 0.1%, respectively. ns, nonsignificant.

N222276.7 ± 0.270.1 ± 0.261.7 ± 1.54.99 ± 0.012.42 ± 0.0124.8 ± 0.59.7 ± 0.11.6 ± 0.40.2 ± 0.1
2877.0 ± 0.369.8 ± 0.360.9 ± 1.05.05 ± 0.012.41 ± 0.0125.8 ± 0.69.7 ± 0.12.1 ± 0.42.1 ± 0.5
Gharib2276.3 ± 0.367.6 ± 0.225.7 ± 2.16.20 ± 0.012.24 ± 0.0116.5 ± 0.112.5 ± 0.135.5 ± 2.359.3 ± 2.0
2874.8 ± 0.264.9 ± 0.420.7 ± 1.46.20 ± 0.012.18 ± 0.0214.9 ± 0.211.9 ± 0.115.4 ± 1.780.9 ± 2.6
LSD 0.7730.8123.5560.0230.0321.120.2714.164.23
Genotype (G)  *** *** *** *** *** *** *** *** ***
Tempe-rature (T)  * *** ns ** ** ns ** *** ***
G × T  ** *** ns ** * * ** *** ***

Biomass, N and NSC partitioning

Although biomass, N and NSC for the different plant parts, including leaves, stem + sheath and panicles, were recorded at different key stages (panicle initiation, flowering, 15 DAF and physiological maturity), we have focused on the data obtained from the last two stages as the former two were unaffected by temperature (Fig. S3). Overall (leaf + stem + panicle) NSC (P < 0.05) and N (P < 0.05) contents in Gharib were reduced significantly at 15 DAF and at physiological maturity, whereas they were relatively unaffected (P > 0.05) in N22 (Fig. 1). A similar pattern was observed with biomass. The percentage NSC content in the panicles at 15 DAF was higher at 28°C than at 22°C in both entries (10% with N22 and 22% with Gharib), with N content following the same trend – N22 and Gharib accumulated 2% and 13% higher N, respectively, with HNT compared with control. However, the exact opposite in NSC and N accumulation in the panicles of both entries exposed to 28°C was recorded during physiological maturity; panicle biomass followed the same pattern. Comparatively, N and NSC contents in the stems in both entries were reduced with HNT at 15 DAF, with the reduction being consistent with stem biomass, whereas the content in the stems was higher with HNT compared with 22°C at physiological maturity.

Figure 1.

Biomass, nitrogen (N) and nonstructural carbohydrate (NSC) partitioning in rice (Oryza sativa) accessions N22 (a) and Gharib (b) at the 15 d after flowering (DAF) and physiological maturity (PM) stages under control and high night temperature (HNT) treatment. Numbers within bars indicate percentage content. NSC content in the leaves was < 5% at both 15 DAF and at physiological maturity in both entries. Values of N, NSC and biomass ‘content’ are obtained from four main tillers of a hill averaged over four replicates from each chamber (i.e. 24 replicate samples).

Rate and duration of grain filling

Using the logistic equation, most of the variation with grain weight during the grain-filling process (R2 = 0.95–0.99) across varieties and temperatures was accounted (Table 3). In N22, the initial, maximum and mean grain-filling rates were increased substantially with HNT compared with the control, whereas it was only the initial grain-filling rate that recorded an increase with Gharib. With HNT, the maximum grain-filling rate was reduced substantially (20.3%) among the spikelets located in the bottom one-third of the panicle in Gharib. Although the mean grain-filling rate increased slightly in spikelets located at the top of the panicle, it was considerably reduced among spikelets located at the middle (2.2%), and particularly in those at the bottom (12.7%), of the panicle compared with the control in Gharib (Fig. S4, Table 3). The time taken to reach the maximum grain-filling rate in N22 was shortened by 1.2–1.7 d across the panicle, whereas, in Gharib, the range was smaller (0.3–0.7 d). The active grain-filling duration in N22 was reduced by 15.6–15.9% under HNT, irrespective of the location of the spikelets on the panicle. Gharib showed a similar response with grain-filling duration, but the effect was much smaller and restricted to the top (4.8%; 0.8 d) and middle (3.2%; 0.6 d) portions of the panicle. The spikelets at the bottom one-third had a much longer grain-filling duration (9.1%; 2 d).

Table 3. Grain-filling parameters for top, middle and bottom portions of the panicle, estimated using logistic equation for rice (Oryza sativa) accessions N22 and Gharib under different night-time temperatures
GenotypePositionTemperature (°C)GR0 (mg per grain d−1)GRmax (mg per grain d−1)GRmean (mg per grain d−1)Tmax (d)D (d) R 2
  1. D, active grain-filling duration; GR0, initial grain-filling rate; GRmax, maximum grain-filling rate; GRmean, mean grain-filling rate; R2, correlation coefficient; Tmax, time taken to reach maximum grain-filling rate.


HNT-responsive flag leaf and spikelet proteins

Two-dimensional gel electrophoresis was carried out on flag leaves and spikelets at two developmental stages, 100% flowering and 12 DAF (EGF), for both tolerant N22 and susceptible Gharib under control and HNT conditions in order to display and compare differentially expressed proteins. Protein profiling revealed c. 400–500 reproducible protein spots using silver staining over a pH range of 4–7 with a molecular weight ranging from 10 to 90 kDa (Figs 2, S5). Protein spots showing significant differential expression in N22 were also identified in Gharib, and their abundance ratio was documented, and vice versa with spots differing from Gharib (Table 4). In addition, the differentially regulated spot expression pattern at a later or earlier stage, depending on the actual sampling stage at either 100% flowering or 12 DAF, respectively, was also recorded to ascertain the developmental stage reprogramming of the proteomes (Table 4). One hundred and three protein spots were differentially regulated in both tissues, including both the developmental stages across N22 and Gharib. Of the 103 spots, 36 spots showing > 1.5-fold change and statistical significance (P < 0.05) in their abundance ratio were excised from N22 gels and analyzed by MS.

Table 4. Abundance ratio (AR = % volume under stress/% volume under control) of differentially expressed protein spots in rice (Oryza sativa) accessions N22 and Gharib at 100% flowering (a) and 12 d after flowering (DAF) (b)
Stage/tissueSpot IDActualReprogrammed
100% flowering (a)AR (N22)AR (Gharib)AR (N22) at 12 DAFAR (Gharib) at 12 DAF
Flag leaf100FL11.884 ± 0.1282.189 ± 0.1396.390 ± 0.2471.752 ± 0.192
100FL31.471 ± 0.3011.893 ± 0.5473.441 ± 0.522 Absent
100FL41.782 ± 0.5081.574 ± 0.1712.264 ± 0.292 Absent
100FL51.484 ± 0.1721.216 ± 0.1242.903 ± 0.1681.499 ± 0.076
100FL60.432 ± 0.0830.349 ± 0.0390.687 ± 0.1320.603 ± 0.098
100FL70.521 ± 0.1240.256 ± 0.2530.688 ± 0.1140.578 ± 0.001
100FL80.490 ± 0.1300.479 ± 0.0190.259 ± 0.0400.157 ± 0.001
100FL110.451 ± 0.1030.537 ± 0.07 Absent Absent
Spikelets100P12.151 ± 0.0281.612 ± 0.1910.077 ± 0.0300.570 ± 0.001
100P32.354 ± 0.7681.633 ± 0.2630.365 ± 0.0320.563 ± 0.087
100P43.004 ± 0.5461.560 ± 0.3382.331 ± 0.1432.380 ± 0.283
100P52.158 ± 0.4881.858 ± 0.1571.538 ± 0.5360.863 ± 0.017
100P62.341 ± 0.2660.890 ± 0.1230.510 ± 0.0051.481 ± 0.004
100P72.163 ± 0.0680.785 ± 0.0920.374 ± 0.1141.594 ± 0.165
100P90.838 ± 0.0261.541 ± 0.091 Absent Absent
12 DAF (b) AR (N22)AR (Gharib)AR (N22) at 100% floweringAR (Gharib) at 100% flowering
  1. In addition, the expression patterns at earlier (spots identified and sequenced at 12 DAF, that is, the early grain filling stage) or later (spots identified and sequenced at the 100% flowering stage) developmental stages were identified and their ARs are presented in italics (reprogrammed). In both cases, bold and normal fonts indicate significant and nonsignificant changes, respectively. Values with ± SE included.

Flag leafEGFL10.289 ± 0.1850.389 ± 0.066 Absent Absent
EGFL20.241 ± 0.1620.479 ± 0.000 Absent 0.436 ± 0.079
EGFL30.316 ± 0.2000.357 ± 0.231 Absent 0.435 ± 0.068
EGFL42.158 ± 0.5301.459 ± 0.0161.508 ± 0.1621.685 ± 0.216
EGFL112.003 ± 0.0291.512 ± 0.2020.604 ± 0.1490.557 ± 0.037
EGFL130.429 ± 0.2810.455 ± 0.107 Absent 2.430 ± 0.011
SpikeletEGFP10.442 ± 0.0400.539 ± 0.025 Absent Absent
EGFP21.657 ± 0.5880.761 ± 0.049 Absent Absent
EGFP31.830 ± 0.6722.829 ± 0.287 Absent Absent
EGFP41.746 ± 0.2171.134 ± 0.150 Absent Absent
EGFP51.898 ± 0.6421.589 ± 0.0270.863 ± 0.0130.818 ± 0.017
EGFP60.272 ± 0.2430.072 ± 0.0321.532 ± 0.0021.978 ± 0.101
EGFP70.355 ± 0.0690.979 ± 0.1020.516 ± 0.0171.979 ± 0.027
EGFP80.135 ± 0.096Absent Absent Absent
EGFP92.153 ± 0.7871.518 ± 0.020 Absent Absent
EGFP100.340 ± 0.0990.493 ± 0.000 Absent Absent
EGFP111.578 ± 0.1471.745 ± 0.200 Absent 1.829 ± 0.142
EGFP122.323 ± 0.7301.063 ± 0.123 Absent Absent
EGFP131.700 ± 0.0401.515 ± 0.424 Absent Absent
EGFP141.485 ± 0.2261.821 ± 0.451 Absent Absent
EGFP152.411 ± 0.9642.090 ± 0.4241.814 ± 0.0590.523 ± 0.047
Figure 2.

Representative two-dimensional gels showing differentially expressed protein spots in rice (Oryza sativa) flag leaves (a, c at 100% flowering and 12 d after flowering (DAF), respectively) and spikelets (b, d at 100% flowering and 12 DAF, respectively) exposed to high night temperatures (HNT) of 28°C. Their actual abundance ratios, obtained from the sampled developmental stage, and their earlier or later stage ratios, depending on the actual sampling stage, are presented in Table 4. The gels shown are from HNT-stressed N22 tissues, and a panel of gels from tissues exposed to control temperature (22°C) is presented in Supporting Information Fig. S5. Gels generated from spikelets at 12 DAF were slightly streaked in both entries, which was mainly caused by excess starch accumulation in the spikelets at the early grain-filling stage.

Among the differentially expressed proteins, 71% and 67% were up-regulated in flag leaves and spikelets of N22 and Gharib, respectively, with all significantly changing spots from the spikelet samples up-regulated in both genotypes at 100% flowering (Table 4a). The same set of spots, when visualized from gels obtained from samples at 12 DAF, showed a clear trend, with the flag leaf spots increasing in intensity and the highly up-regulated spikelet spots down-regulated in N22. The pattern was not clear with the susceptible Gharib. Only 7% and 25% of flag leaf and spikelet spots, respectively, identified at the 100% flowering stage, were not detected in gels at the 12-DAF stage (Table 4a). From tissues obtained from the 12-DAF stage, that is coinciding with the EGF stage, 42% and 56% of the significantly changing spots were up-regulated in both flag leaves and spikelets of N22 and Gharib, respectively (Table 4b). The direction of change with both the up- and down-regulation of spots was identical in both N22 and Gharib, whereas the intensity of change in both directions differed, being stronger in N22 across all significantly changing spots, except for EGFP3, 6, 11 and 14. An examination of spots at the earlier developmental stage (100% flowering), which were actually extracted and sequenced at 12 DAF, revealed that 75% and 44% of the spots were undetected in N22 and Gharib, respectively.

In total, 36 differentially regulated proteins in response to HNT stress were grouped into seven categories according to their putative physiological functions: heat shock proteins (HSPs) and other molecular chaperones; proteins involved in signaling; proteins involved in sugar metabolism; proteins involved in nucleic acid/protein modification and repair; ribosomal proteins; proteins involved in phytohormone biosynthesis and signaling; and others (Table 5). Specific HSPs, proteins involved in calcium signaling and in nucleic acid/protein modification and repair were highly up-regulated in the case of N22 (as compared with Gharib) in response to HNT stress. Proteins involved in photosynthesis were down-regulated in both varieties (Table 5).

Table 5. Differentially expressed proteins identified by matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF MS) and their annotation derived from MASCOT or PROFOUND databases, together with their experimental and theoretical pI and molecular weight (Mr)
Spot IDIdentified protein descriptionFold change (actual)Fold change (reprogrammed)
Protein nameLocus IDMSTheoreticalExperimentalMP/PSSC%N22GharibN22Gharib
  1. Protein spot numbers are as indicated in Fig. 2. Corresponding accession numbers can be obtained from the TIGR database ( Actual and reprogrammed fold change as defined in Table 4.

  2. +, ++, +++, up-regulation by 1.25–1.49-, 1.5–1.9-, ≥ 2.0-fold; -, –, —, 1.25–1.49- (0.87–0.75 abundance ratio (AR)), 1.5–1.9- (0.74–0.50 AR), ≥ 2.0 (≤ 0.49 AR)-fold down-regulation. Ab, absent; CUE, coupling of ubiquitin to ER degradation; GCN5, general control non-repressed protein 5; IAA, indole-3 acetic acid; HSP, heat shock protein; MP/PS, matched peptides/searched peptides; MS, MASCOT score; PQ, total peptide searched per query; SC, sequence coverage.

HSPs and other chaperones
EGFP2Peptidyl-prolyl cistrans isomerase, FKBP-typeLOC_Os06g20320.1425.1951.95.16518/4220++-AbAb
EGFP13Hsp20/α crystalline family proteinLOC_Os01g08860.1537.8517.16.7648/1042++++AbAb
100FL 1DnaK family proteinLOC_Os01g08560.2335.1592.54.862.410/1513++++++++++
EGFP11Late embryogenesis abundant proteinLOC_Os05g46480.4 9.116.556.8125/589++++Ab++
Proteins involved in signaling
EGFP4CAMK_CAMK_like.17 – CAMK includes calcium/calmodulin-dependent protein kinasesLOC_Os03g03660.4365.5061.66.96217/3915++=AbAb
100P7Calmodulin-binding proteinLOC_Os02g08120.2
EGFL11IQ calmodulin-binding motif family proteinLOC_Os05g43670.1496.4662.15.160.811/1313+++++
100FL3Phosphatidylinositol 3- and 4-kinase family proteinLOC_Os06g17290.1 5.596.85.470.214/2215++++++Ab
Proteins involved in sugar metabolism
EGFP1Hydrolase, α/β fold family proteinLOC_Os03g17900.1505.7358.85.067.514/2028AbAb
EGFP5NAD-dependent epimerase/dehydratase family protein/UDP-glucose 4-epimeraseLOC_Os05g51670.1345.5438.55.8454/511++++--
EGFP12β-Mannosidase/glucosidase homologLOC_Os01g32364.1446.7757.46.8645/612+++=AbAb
EGFP15UTP–glucose-1-phosphate uridylyltransferaseLOC_Os09g38030.1645.6753.55.5409/917++++++++
100P1Ribose-5-phosphate isomerase ALOC_Os07g08030.1645.7229.44.345.39/1237+++++
Proteins involved in nucleic acid/protein modification and repair
EGFP7OsPDIL2-1 protein disulfide isomerase PDIL2-1,LOC_Os05g06430.2465.4440.94.53015/4138=++
100FL4CUE degradation domain containing proteinLOC_Os03g10750.1 5.996.85.469.214/1715+++++++Ab
100FL5Histone acetyltransferase GCN5LOC_Os10g28040.1666.0165.54.54214/1819+=++++
100P6Maturase KLOC_Os04g16734.1439.64625.476.334/3435+++=+
100P4Retrotransposon protein, putative, Ty3-gypsy subclassLOC_Os04g48660.1 5.793.25.178.216/2616+++++++++++
EGFP3Retrotransposon protein, putative, unclassifiedLOC_Os03g28020.1435.7461.96.86220/4517+++++AbAb
Ribosomal proteins
EGFP860S ribosome subunit biogenesis protein NIP7LOC_Os07g47420.1429.021.16.2206/639AbAbAb
EGFP9Chloroplast 30S ribosomal protein S8LOC_Os04g16832.16410.7166.3188/1152+++++AbAb
Proteins involved in phytohormone biosynthesis and signaling
EGFP14Auxin-responsive Aux/IAA gene family memberLOC_Os02g57250.2435.8226.74.6805/531+++AbAb
EGFL2Gibberellin 2-β-dioxygenase 7LOC_Os04g44150.1 4.821.25.312.210/1645Ab
EGFL4Oxidoreductase, short-chain dehydrogenase/reductase family domain-containing proteinLOC_Os04g44920.3 4.921.094.412.06/2136++++++++
EGFL13Cytokinin-N-glucosyltransferase 1LOC_Os03g60960.1588.6451.006.5307/917Ab+++
100FL6Ribulose bisphosphate carboxylase small-chain, chloroplast precursorLOC_Os12g17600.1669.0419.96.314.29/7040
EGFL3Ribulose bisphosphate carboxylase large-chain precursorLOC_Os10g21268.1646.3529.115.7126/2421Ab
EGFP6KUN1 - Kunitz-type trypsin inhibitor precursorLOC_Os04g44470.1384.9720.04.2155/1036++++
EGFP10Cation efflux family proteinLOC_Os05g03780.1526.9214.86.7204/626AbAb
100FL7Homeobox and START domains-containing proteinLOC_Os01g55549.1636.6345.46.41213/2129
100FL8Coiled-coil domain-containing protein 25LOC_Os05g13440.1 5.625.55.6137/3222
100FL11Ankyrin repeat family proteinLOC_Os05g01310.1
100P5DUF630/DUF632 domains-containing proteinLOC_Os08g43730.1
100P9Stromal membrane-associated proteinLOC_Os06g40704.2426.1235.975.533.813/2924-++AbAb
100P3Expressed proteinLOC_Os01g34430.1595.7157.224.737.835/5246+++++
EGFL1Expressed proteinLOC_Os08g08930.1 4.812.184.811.22/951AbAb


Conclusions drawn from controlled environment experiments have documented an HNT-induced increase in respiration rate and decrease in pollen germination (Mohammed & Tarpley, 2010) and poor assimilate translocation to grains (Morita et al., 2005; Cheng et al., 2009; Kanno & Makino, 2010), with a subsequent reduction in seed set and/or grain weight. These conclusions are based on individual genotype performance, whereas our study builds on the outcome of a wide genetic diversity screening (36 accessions) and tests the most contrasting entries from both studies using the same chambers established under field conditions.

A moderate increase in night temperature during the entire reproductive period led to a significant decline in grain yield and total dry matter at physiological maturity with the highly sensitive Gharib. This decline in yield was mainly attributed to a substantial reduction in 1000-grain weight, a phenomenon observed by Morita et al. (2005) and Kanno & Makino (2010). However, the percentage seed set was unaffected in Gharib, which contrasted with the finding of Mohammed & Tarpley (2009a), who noted a 90% reduction in fertility at HNT of 32°C using cultivar Cocodrie (which could be highly susceptible). Our ongoing controlled environment work indicates a similar response from very sensitive varieties exposed to temperatures > 30°C, but the tolerant N22, even under 35°C HNT, recorded a < 5% reduction in sterility (O. Coast et al., unpublished; University of Reading, UK). Hence, preliminary diversity analysis is essential to avoid an overestimation of the temperature effects. In addition, the grain weight of tolerant N22 was unaffected in our field study. However, the number of spikelets per panicle was reduced significantly in N22 with HNT, accompanied by a higher seed set, demonstrating the plastic response of maintaining yield under HNT. Competition for assimilates between the spikelets and the stem during panicle formation has been documented, with spikelets being poorer competitors than the stem for available assimilates (Fischer & Stockman, 1980). In this competition for assimilates between panicle and stem, the stem in N22 appears to have prevailed over the panicle, as evidenced by a 6.5% increase in height and a simultaneous decrease (14.6%) in spikelet number under HNT (Table 1). Moreover, a similar quantitative impact of HNT on spikelet degradation (9.6%) in N22 was observed in an independent experiment using the same chambers, but such plastic responses were not observed with the sensitive Gharib.

A steady supply of assimilates in the 010 and 1020 d following heading is a crucial determining factor for endosperm expansion and grain filling, respectively (Nagata et al., 2001). Carbohydrates for grain filling could either be assimilated during the ripening period or translocated from assimilates accumulated in the leaf sheath and culms before heading (Nagata et al., 2001; Lafarge & Bueno, 2009). In our study, a significant decline in N and NSC content in the sensitive Gharib throughout the ripening period until physiological maturity resulted in assimilate shortage and, with reduced 1000-grain weight and grain yield, indicated a greater limitation with source, although sink strength reduction could not be ruled out. After accounting for the accumulated N after flowering from the initial content + the translocation from the leaves and stem, unaccounted values of 11 mg per hill and 26.2 mg per hill N were recorded in N22 and Gharib panicles, respectively, at 22°C, and 14 mg per hill in both entries at 28°C, indicating the contribution of direct N uptake or active translocation of N stored in the roots during the active grain-filling stage (Fig. 3). Compared with N, NSC translocation to the panicle was more pronounced, with a higher contribution from stem NSC than from leaf NSC (data not shown), as documented earlier (Fu et al., 2011).Comparison of N22 and Gharib across both temperatures independently showed a smaller decrease in NSC translocation in Gharib, which could be equated to the HNT effect only, whereas a larger decrease in N22 could be caused by a combination of the HNT effect and reduced sink size (Fig. 3).

Figure 3.

Remobilization of nitrogen (N; mg per hill) from rice (Oryza sativa) leaves (N content in leaves at flowering – N content in leaves at physiological maturity (PM)), stem (N content in stems at flowering – N content in stems at PM) and possibly roots into panicles, that is, N increase in panicles (N content in panicles at PM – N content in panicles at flowering), and nonstructural carbohydrate (NSC) translocation ratio from stem to panicles (amount of NSC transferred from stem to grains/NSC in stems at flowering × 100) of N22 (a) and Gharib (b); white circles represent 28°C and black circles represent 22°C. Horizontal bars indicate ± SE.

Grain filling, the final stage of growth in cereals, is determined by the product of the rate and duration of grain growth. A negative relationship between the rate and duration of grain filling has been established (Yang et al., 2008). N22, which has considerably higher initial, maximum and mean grain-filling rates across the whole panicle, was able to compensate for a significant reduction in active grain-filling duration and maintained grain yield (Table 3). Interestingly, the plastic behavior of N22 to the deliberate reduction in the number of spikelets per square meter probably allowed the remaining spikelets to receive sufficient assimilates within the shortened grain-filling duration, a response that was absent in the susceptible Gharib. In addition, this response would allow assimilate saving, which otherwise would have been utilized for the production of additional nonproductive spikelets. Gharib, however, showed a higher initial grain-filling rate, but the maximum and mean grain-filling rates were decreased greatly, in both middle and bottom portions of the panicles, together with the grain-filling duration in the top and middle parts of the panicle, thereby reducing the final grain weight. Our results confirm the conclusions of Kobata & Uemuki (2004) that a lower yield caused by high temperature during grain filling may be a result of the failure of assimilate supply to meet the accelerated grain-filling rate. This was the case with Gharib. Further, a significant synergistic correlation between the grain-filling rate and grain weight (but not between the grain-filling duration and grain weight) in bread and durum wheat under high temperature has been recorded (Dias & Lidon, 2009). Ideally, rice varieties with sufficient biomass, equipped with efficient translocation efficiency (high grain-filling rates) to compensate for the reduced grain-filling period, could potentially overcome the impact of HNT on grain yield.

HSPs are functionally involved in the repair and renaturation of stress-damaged proteins, in addition to protecting the cells against the effects of stress (Wang et al., 2004; Sarkar et al., 2009; Jagadish et al., 2011). Peptidyl-prolyl cistrans isomerase (FKBP-type) was particularly up-regulated in early grain-filling spikelets in N22 and was down-regulated in the case of Gharib, with PPIase (peptidylprolyl isomerase) possibly having a positive role in maintaining protein synthesis and trafficking proteins during the active grain-filling stage. This protein is known to be induced in floral tissues under heat stress in wheat (Kurek et al., 1999) and works in tandem with HSP90 to ensure the correct folding of proteins in Arabidopsis thaliana (Hagai et al., 2007). Late embryogenesis abundant protein, which behaves like HSP12 in Saccharomyces cerevisiae, was up-regulated in the early grain-filling panicle of both varieties, showing its role in grain filling under heat stress and preventing other proteins from heat-induced desiccation. Calcium, a universal signaling molecule under heat stress, triggers cytosolic Ca2+ bursts, which are transduced by several Ca2+-binding proteins (CBPs), such as calmodulin (CaM), CaM-related proteins, Ca2+-dependent protein kinases (CDPKs), etc., that further up-regulate the expression of HSPs (Liu et al., 2003; Yang & Poovaiah, 2003). In our study, CBPs, such as CaM-dependent protein kinases, CaM-binding protein and IQ CaM-binding motif family protein, were more strongly up-regulated in tolerant N22, whereas the first two proteins were unchanged and down-regulated, respectively, in the susceptible Gharib panicles. Phosphatidylinositol 3- and 4-kinase family protein, which is involved in phosphate signaling in animals, was up-regulated at the 100% flowering stage, but more strongly at 12 DAF, indicating its role in high temperature stress signaling in N22, whereas the same protein was undetected in susceptible Gharib. Among the proteins involved in sugar metabolism, β-mannosidase/glucosidase homolog was highly up-regulated only in N22, whereas the three other proteins were equally up-regulated in both entries. The CUE (coupling of ubiquitin to ER degradation) domain-containing protein, which is involved in the degradation of misfolded proteins in the endoplasmic reticulum and protein sorting, was up-regulated in both varieties, with a higher level of expression in N22 at the EGF stage. In addition, histone acylation by GCN5 (general control non-repressed protein 5) and HAC (histone acetyl transferase) helps in the transcriptional regulation of HSP70 and HSP17 genes, which are actively involved in correct protein folding and sequestration under high temperature stress (Bharti et al., 2004; Han et al., 2008). Maturase K could assist in splicing its own and other chloroplast group II introns, showing more active transcription of heat stress-responsive gene up-regulation in N22 (but down-regulation in Gharib). Proteins involved in the biosynthesis of RuBISCo were down-regulated in both genotypes, which could result in reduced photosynthetic rate with a pre-exposure to HNT, a phenomenon documented in wheat (Prasad et al., 2008). The majority of the significantly changing proteins at the 100% flowering stage were detected at 12 DAF in both flag leaves and spikelets, whereas those that were sequenced from tissues at 12 DAF were undetected at 100% flowering. This indicated dynamic proteome programming with different tissues at key developmental stages in rice when exposed to HNT. The combined increase in HSPs and Ca signaling proteins, and the better nucleic acid/protein modification and repair in tolerant N22 at the EGF stage, could have allowed for better enzymatic activity in the conversion of sucrose to starch.

Rice market prices are largely determined by milling quality outcomes and appearance, that is, higher chalk or brokens reduce rice prices dramatically. The significant reduction in milled rice yield and the increase in chalk content (with the highest chalk category in Gharib) are proxy for the negative impact of HNT on grain weight (reduced grain width), leading to reduced yield and total milled rice. The decrease in grain width could be associated with a reduction in average endosperm cell area observed under HNT (Morita et al., 2005), or with abnormal amyloplast packaging, resulting in white core chalk formation (Ishimaru et al., 2009). From source–sink manipulation studies, a close relationship between assimilate supply and milky white chalk formation has been established (Tsukaguchi & Iida, 2008), with increasing assimilate supply overcoming chalk formation even under high temperatures (Kobata & Uemuki, 2004). In addition, higher maintenance respiration with increasing night temperatures could partly be responsible for reduced assimilate supply, as documented by Cheng et al. (2009) and Mohammed & Tarpley (2010). Chalkiness was not a problem with N22, mainly because of the increased grain-filling rates and little influence on overall biomass, even under HNT. Interestingly, chalkiness under the 5075% category was reduced significantly in Gharib with HNT, a feature that could be attributed to better assimilate transfer at the initial grain-filling stages, but, with a lack of sustained supply of assimilates, this resulted in a 36.4% increase in the > 75% chalkiness category. Moreover, Gharib with a comparatively higher biomass than N22 could have a relatively higher demand for maintenance respiration, depriving a larger share of assimilates over the 2-month-long HNT exposure.


This research was funded by the Federal Ministry for Economic Cooperation and Development, Germany and United States Agency For International Development – Bill and Melinda Gates Foundation-BMGF (Cereal Systems Initiative for South Asia). We also thank the GRiSS scholarship program (through GRiSP – Global Rice Science Partnership) for the MS graduate scholarship to the first author. We are also grateful to our technical staff who helped carry out the research. We deeply appreciate the initial guidance of Dr Laza in establishing the experiment. The authors declare no conflict of interest.