Introduction

The Intergovernmental Panel on Climate Change (IPCC 2007) has developed future climate scenarios based on expected increases in temperature and atmospheric concentrations of greenhouse gases such as carbon dioxide (CO2), methane, and nitrous oxide. Increased concentrations of CO2, coupled with higher temperatures, are likely to have a considerable impact on the future dynamics of water and nutrients, and also on the structure and function of ecosystems (Cramer et al. 2001; Henry et al. 2005). This, in turn, will lead to substantial changes in agricultural crops (Peng et al. 2004; Yoshimoto et al. 2005; Masutomi et al. 2010), which are widely distributed and grown using the same agricultural management practices.

The Japanese archipelago extends over a distance of approximately 2,000 km, and its climate ranges from subtropical to subarctic (Japan Meteorological Agency 2005). There are concerns that climate change in the region will affect not only the production but also the quality of rice (Oryza sativa L.)—one of the most important basic food resources in Japan. The impact of climate change on rice production has been investigated extensively using crop models and climate change scenarios (Matthews et al. 1997; Aggarwal and Mall 2002; Tao et al. 2008). The results suggest that global warming will enhance rice production in northern Japan, but reduce production by at least 15–30 % in the south-central and southwestern parts of the country (Horie et al. 1995). In addition, high temperatures during the grain-filling period may reduce rice quality, as measured by grain size, transparency, roundness, and cracking (Yamakawa et al. 2007; Morita 2008; Hasegawa et al. 2011). However, the way in which high air temperature, especially at night, decreases the fertility of rice grains remains unclear (Cheng et al. 2009).

Carbon isotope discrimination (Δ13C) occurs in C3 plants, owing to series of isotope fractionations along the photosynthesis pathway. As a consequence, plant carbon is invariably depleted in 13C in respect to the isotopic composition of atmospheric CO2. The parameter Δ13C is an important indicator of all those conditions affecting the photosynthetic set point, i.e., the internal to atmospheric CO2 concentration ratio (Ci/Ca) (Farquhar et al. 1989). The δ13C of plants depends on fractionation during diffusion of CO2 into the leaf, and also on subsequent photosynthetic metabolism and water-use efficiency (WUE) independently (Farquhar et al. 1982; Kume et al. 2003; Hanba et al. 2010; Ma et al. 2012). Most carbon stored in mature rice grains originates from CO2 assimilation during the grain filling period, with the flag leaf as the most photosynthetically active (Murchie et al. 1999). Factors such as the photosynthesis rate of the flag leaf during this period could potentially determine grain yield (Dingkuhn et al. 1989). Furthermore, the plastic response of WUE to water availability has been demonstrated using instantaneous gas-exchange measurements, by analyses of plant δ13C (Ehleringer and Cooper 1988; Toft et al. 1989; Centritto et al. 2009). Therefore, rice δ13C may reflect an integration over time of photosynthesis and plant water status (with related factors such as sunshine hours, precipitation, humidity, and air temperature), as this is affected by water availability and evapotranspiration demand by the atmosphere.

Oxygen isotope discrimination (Δ18O) in C3 plants occurs mainly during transpiration at the leaf relating to the atmospheric to internal vapor pressure ratio (Barbour 2007). Although the oxygen stable isotopic composition (δ18O) of xylem water reflects purely soil water taken up by roots (Dawson and Ehleringer 1991), there are complex phenomena of remixing and back diffusion of the enriched water with xylem water (i.e., metabolic water) arriving to the leaves and permeating the chloroplasts, consequently determining the δ18O of metabolic water (Farquhar and Lloyd 1993). Additionally, the δ18O of soil water is normally closely linked to precipitation, which becomes enhanced with δ18O as the temperature increases at lower latitudes and altitudes (Dansgaard 1964; Mizota and Kusakabe 1994). The metabolic water retains the environmental signal (both soil water and transpiration effects) and is detectably affected by air temperature, which imprints the isotopic signature of the organic matter by means of isotopic exchange on the gem-diole groups of triose-phosphates (Samuel and Silver 1965; Sternberg et al. 1986; Roden and Farquhar 2012). Maximum air temperature (primarily during the day) has the potential to affect both photosynthesis and transpiration, whereas minimum air temperature (primarily at night) may affect transpiration, which promotes translocation in mature rice grains. Therefore, it is expected that the δ13C and δ18O of organic matter in rice grains will reflect physiological changes in response to warming impacts. If the δ13C and δ18O of rice grains reflect an increase in maximum and/or minimum air temperatures, they may provide direct indicators for changes in rice yield and quality under conditions of climate change.

The aim of the present study was to investigate whether stable carbon and oxygen isotopes of rice organic matter reflect variations in air temperature during grain filling, and are related to the yield and quality of rice. We analyzed the δ13C and δ18O of rice grains from the northern latitudes (30.49°–37.14°) of Japan in relation to mean maximum and minimum air temperatures, yield, and quality—as defined by the content of first-grade rice.

Materials and methods

Rice and rice-field water samples

A total of 205 samples of the japonica rice cultivar Koshihikari (Oryza sativa L.) were obtained from 114 paddies within 21 prefectures in September 2007, and 91 paddies within 13 prefectures in September 2008, on the Japanese islands of Honshu, Shikoku, and Kyusyu (detailed in Fig. 1; Table 1). We selected and polished rice crops for which heading and flowering occurred between July and August. Each polished rice grain sample (50 g) was freeze-dried and ground into a powder for stable isotope analysis. In May 2007, water samples were filtered by a membrane filter (nominal pore size: 0.45 μm, ADVANTEC, Tokyo, Japan) and collected in 100 ml polyethylene bottles from the irrigation ditch of one rice field in each of the following locations (within the region 24.35–38.17°N): Sado (Niigata), Uonuma (Niigata), Murakami (Niigata), Matsusaka (Mie), Matsukawa (Nagano), Matsumoto (Nagano), Matsue (Shimane), Unnan (Shimane), and Ishigaki (Okinawa) in order to consider geographical changes in isotopic composition of source water (Fig. 1). The rice crops were irrigated and fertilized according to conventional management practices for rice farming in Japan. Intermittent irrigation was applied from approximately 1 month after transplanting to maintain wet soil conditions. Irrigation was stopped from approximately 1 month after ear emergence until harvest, because precipitation water provided sufficient water levels for crop production (e.g., Chapagain and Yamaji 2010).

Fig. 1
figure 1

Sampling locations of Koshihikari rice (Oryza sativa L.) and rice-field water in Japan. Closed and open circles represent locations of rice samples in 2007 and 2008, respectively. Open triangles represent locations of rice-field water samples in 2007

Full size image
Table 1 Yield, proportion of first-grade rice, δ13C, and δ18O of Koshihikari rice (Oryza sativa L.), and sunshine hours, precipitation, and mean maximum and minimum air temperatures during grain filling in 2007 and 2008
Full size table

Meteorological and rice statistical data collection

Meteorological data (mean maximum and minimum air temperatures, precipitation, and sunshine hours during the grain-filling period) were obtained from local meteorological stations within a 15-km radius of each sampling point (Japan Meteorological Agency 2012). Where possible, these data were collected for the grain-filling period of Koshihikari rice, determined according to statistical reports (Ministry of Agriculture, Forestry and Fisheries 2012). Rice yield and grade data for 2007 and 2008 were collected from statistical reports (Ministry of Agriculture, Forestry and Fisheries 2012). We used the characteristics of first-grade rice (<15 % transparency, roundness, and cracking) as indicators of quality.

Stable isotope analysis

The δ13C of rice grain (δ13Crice) was determined using an elemental analyzer/isotope ratio mass spectrometer (EA/IRMS, Finnigan Delta V Advantage, interfaced with FlashEA 1112 HT; Thermo Fisher Scientific, Bremen, Germany) (Suzuki et al. 2008). Subsamples of rice (1.0 mg) were placed into tin capsules (5 × 9 mm) and loaded into the auto-sampler of the EA/IRMS. The obtained CO2 was separated on a PorapakQS column (length, 3 m) at 40 °C. The δ notation defined in Eq. (1) was used to describe the isotopic composition.

$$ \delta (\permille) = (R_{\rm{sample}} /R_{\rm{standard}} - 1) \times 1000 $$
(1)

where R sample is the isotope ratio of the sample, and R standard is the isotope ratio of the international standard, Vienna Pee Dee Belemnite. Each sample was determined in duplicate, and analytical errors in δ13C had a standard deviation (SD) <0.1 ‰.

The δ18O of rice grain (δ18Orice) and rice-field water (δ18Owater) were determined by EA/IRMS in the pyrolysis mode. Subsamples of rice and rice-field water (0.5 mg and 1.0 μL, respectively) were placed into silver capsules (5 × 3 mm). Care was taken to avoid atmospheric contamination of the samples. All rice samples were dried overnight by lyophilization before isotope measurement. Pyrolysis was performed in a ceramic tube of glassy carbon at 1,450 °C under a continuous flow of He at 100 ml min−1. The gas obtained was separated from residual gases by molecular sieving through a 5-A˚ column (length, 3 m) at 50 °C (Werner et al. 1996; Böhlke et al. 2003; Leuenberger and Filot 2007). δ18O was calculated according to Eq. (1) and expressed relative to that of Vienna Standard Mean Ocean Water (V-SMOW). The measured δ18O values were obtained using known isotope benzoic acid standards (+71.4, +23.2 ‰) purchased from Indiana University (Indianapolis, IN). The accuracy of the obtained δ18O values was checked independently against V-SMOW and Vienna Standard Light Antarctic Precipitation standards distributed by the International Atomic Energy Agency. Four working standards—dibenzo-24-crown-8 (−15.7 ‰), dibenzo-18-crown-6 (+1.7 ‰), β-d-galactose pentaacetate (+12.7 ‰), and D-(+)-sucrose octaacetate (+26.8 ‰)—were determined every six samples to confirm the reproducibility of the measurements (Nakashita et al. 2008). The analytical error of the isotope measurements had a SD <0.3 ‰, with a minimum sample amount of 10 μmol oxygen.

Statistical analysis

Correlations between rice (δ13C, δ18O, yield, and quality) and meteorological data (mean maximum and minimum air temperatures, precipitation, and sunshine hours during the grain-filling period) were tested independently using Pearson’s product-moment coefficient with the average values for each prefecture. We excluded from our statistical analyses the rice yield and quality of 2007 data from Kagoshima and Miyazaki prefectures, because of typhoon damage. Correlations between the δ18Owater and the δ18Orice, and also between the mean minimum air temperature during grain filling and the δ18Orice and the δ18Owater with global mean 18O enrichment of carbohydrate relative to the source water (+27 ‰, Sternberg 1989), were tested using Pearson’s product-moment coefficient. In this study, the 18O enrichment of rice grain to source water was defined as δ18Orice–δ18Owater (e.g. Matsuo et al. 2013). We compared the theoretical values (δ18Owater + global mean 18O enrichment) with measured δ18Orice, because the difference between the theoretical and measured 18O enrichments may reveal physiological changes such as rice translocation and evaporative demands. For all tests, an α value of 0.05 was used to indicate statistical significance. All analyses were conducted using R ver. 2.15.0 (R Development Core Team 2012).

Results

Rice yields ranged from 2.05 to 6.27 t ha−1 in 2007, and from 4.67 to 6.35 t ha−1 in 2008 (Table 1). The proportion of first-grade rice in the rice yield (rice quality) ranged from 0 to 97 % in 2007, and from 30 % to 98 % in 2008. There was a positive correlation between yield and quality (Table 2). Sunshine hours during grain filling ranged from 144 to 276 h in 2007, and from 118 to 268 h in 2008. Sunshine hours were not correlated with yield or quality of rice (Table 2). Precipitation during grain filling ranged from 10 to 701 mm in 2007, and from 37 to 389 mm in 2008. Precipitation was also not correlated with yield or quality of rice (Table 2).

Table 2 Results of Pearson’s product–moment coefficient with the average values for each prefecture between rice grain (yield, quality, δ13C, and δ18O) and meteorological data (sunshine hours, precipitation, and maximum and minimum air temperatures)
Full size table

Rice δ13C ranged from −28.4 to −26.8 ‰ in 2007, and from −28.1 to −27.1 ‰ in 2008 (Table 1). Rice δ13C was not correlated with yield, quality, sunshine hours, or precipitation (Table 2). δ18Orice ranged from 20.8 to 24.1 ‰ in 2007, and from 20.4 to 23.9 ‰ in 2008 (Table 1). δ18Orice was not correlated with sunshine hours, yield, quality or precipitation (Table 2). The δ18Orice was not correlated with the δ13C (Table 2).

The mean maximum air temperatures ranged from 30.6 to 34.5 °C in 2007, and from 28.3 to 34.5 °C in 2008 (Table 1). The maximum air temperature was negatively correlated with rice yield (Fig. 2a), quality (Fig. 2b), but not with δ13C (Fig. 2c), or δ18Orice (Fig. 2d).

Fig. 2
figure 2

Relationships among mean maximum and minimum air temperatures during grain filling, and yield, quality, δ13C, and δ18O of Koshihikari rice (Oryza sativa L.). a Yield versus mean maximum air temperature, b quality versus mean maximum air temperature, c δ13C versus mean maximum air temperature, d δ18O versus mean maximum air temperature, e yield versus mean minimum air temperature, f quality versus mean minimum air temperature, g δ13C versus mean minimum air temperature, h δ18O versus mean minimum air temperature. Data are represented as mean ± SD

Full size image

The mean minimum air temperatures ranged from 19.4 to 24.9 °C in 2007, and from 19.3 to 24.6 °C in 2008 (Table 1). The minimum air temperature was negatively correlated with rice yield (Fig. 2e) and quality (Fig. 2f), positively correlated with δ18Orice (Fig. 2h), but not correlated with rice δ13C (Fig. 2g).

There was a positive correlation between the δ18Owater and δ18Orice (Fig. 3). The 18O enrichment of rice relative to rice-field water decreased significantly with an increase in the mean minimum air temperature during grain filling (Fig. 4).

Fig. 3
figure 3

Relationship between the δ18O of rice-field water and Koshihikari rice (Oryza sativa L.) in different regions of Japan

Full size image
Fig. 4
figure 4

Relationship between mean minimum air temperature during grain filling and the δ18O of Koshihikari rice (Oryza sativa L.) and rice-field water with global mean 18O enrichment of carbohydrate relative to the source water (+27 ‰)

Full size image

Discussion

In the present study, the δ18Orice was correlated positively with the δ18Owater (Fig. 3), confirming that the oxygen atoms in the organic compounds of rice grain were derived mainly from the region-specific ambient water. The δ18O of precipitation and groundwater (i.e., source water for rice plants) varied in accordance with climatic and geographical changes relating to amount and thermal effects (Dansgaard 1964; Rozanski et al. 1992), and becomes enhanced at lower latitudes and altitudes in Japan (e.g., Mizota and Kusakabe 1994). Therefore, similar to variations in the δ18O of organic compounds (e.g., cellulose, sucrose, etc.) in trees to climate gradients in air temperature, rainfall, sunshine hours, and water availability during the growing season (Libby and Pandolfi 1974; Cernusak et al. 2003; Li et al. 2011), the δ18Orice may reflect temperature-induced changes in the δ18Owater. However, the δ18O of irrigation water may not fully reflect precipitation δ18O in some locations, because irrigation water is obtained from various sources—not only rivers, but also ground water and springs far away from the paddy fields. For instance, water used in paddy fields may be derived from precipitation in the mountains either directly or via ground water or springs. As a result, sources of water used in different paddy field regions need to be examined in order to reflect temperature-induced changes.

We found negative effects of mean minimum air temperature during grain filling on the δ18Orice (Fig. 2h), but no significant effects of mean maximum air temperature (Fig. 2d). Throughout many parts of Japan, global warming is predicted to lead to a greater increase in minimum air temperatures than in maximum air temperatures (Japan Meteorological Agency 2005). Therefore, the δ18Orice is potentially a good indicator for climate-change-induced variations in minimum air temperature. The mean enrichment of δ18Orice relative to source water was +31 ± 2 ‰, it is 4 ‰ higher than the global mean 18O enrichment of carbohydrate relative to source water of +27 ± 4 ‰ (Sternberg 1989). The oxygen isotope enrichment of rice grain decreased as the minimum air temperature increased and the difference between measured and global mean 18O enrichments decreased with increase in the minimum temperature (Fig. 4). A rise in minimum air temperature forecast to cause increase in respiration and translocation (Morita 2008), having the potential for increase in δ18O of the rice grain with degradation of starch. Additionally, the increase of minimum air temperature may induce an increase in soil and metabolic water temperatures, and may act to promote of smaller isotope enrichment during transpiration and photosynthesis compared to low water temperatures, because high water temperature increases stomatal conductance and leaf photosynthesis (Shimono et al. 2004). In this study, although we did not accumulate data for other environmental factors such as humidity, altitudes, and nutrients, they may help to demonstrate the effects of minimum temperature on δ18Orice. The δ18O of rice carbohydrate is also dependent on the actual metabolic status of plants as reported by Saurer et al. (1997).

Many physiological studies have indicated that decreases in grain size and quality, such as transparency, roundness, and cracking, are caused by abnormal growth of the endosperm under high temperatures (e.g., Morita 2008; Hasegawa et al. 2011). For example, rice grain size was shown to decrease when night air temperatures exceeded 21 °C during grain filling (Matsushima and Tsunoda 1957; Aimi et al. 1959). In the present study, we detected stronger negative effects from mean minimum air temperature during grain filling on rice yield and quality than from mean maximum air temperature (Fig. 2). Yamakawa et al. (2007) reported that several starch synthesis–related genes, such as granule-bound starch synthase I (GBSSI), branching enzymes (especially BEIIb), and a cytosolic pyruvate orthophosphate dikinase, are downregulated by high temperatures, whereas genes for starch-consuming α-amylases and heat shock proteins are upregulated. Therefore, an increase in minimum air temperature during grain filling can cause changes in the processes involved in amylose composition and translocation, and might lead to changes in the δ18Orice. Minimum air temperature could therefore be useful for predicting the yield and quality of rice, with changes in the δ18Orice reflecting the physiological conditions during grain filling.

In contrast to the δ18Orice, the δ13Crice was not correlated with minimum or maximum air temperatures, suggesting that the δ13Crice might not be a direct indicator of changes in air temperature. The δ13C of rice is known to be a good indicator of WUE (Kondo et al. 2004; Xu et al. 2009). In the present study, we observed a relatively small difference in rice δ13C among the prefectures surveyed (1.6 ‰), suggesting nearly constant water-use efficiency. The δ13Crice might reflect the nearly constant wet soil conditions maintained under the unified irrigation and fertilization practices of conventional rice farming in Japan, irrespective of the differences in plant-experienced atmospheric vapor demand, namely the leaf-to-air vapor pressure difference caused by different atmospheric temperatures, hours of sunshine, and precipitation. We determined no effects of precipitation on the yield and quality of rice; this finding could also be derived from cultivation under constantly wet soil conditions.