Photosynthesis, Grain Yield, and Nitrogen Utilization in Rice and Wheat

Rice (Oryza sativa) and wheat (Triticum aestivum) are the two most commercially important crops, accounting for more than 40% of global food production. They were domesticated in different climates and differ largely in their growth environments: Rice is tropically cultivated in hot, wet climates, whereas wheat tends to be grown in cooler temperate climates. However, both crops have been bred in similar directions. For example, the introduction of semidwarf traits into rice and wheat from Chinese and Japanese varieties in the 1960s made a great contribution to increasing yield in both species (Evans, 1997). Since the semidwarf cultivars can use large inputs of nitrogen (N) fertilizer without lodging, the introduction of dwarfing genes allowed the production of varieties with high leaf N content and enhanced sink capacity. Large inputs of N fertilizer in turn have drawn much attention to the environmental impact of N fertilization practices (Cassman et al., 1998). Therefore, it is important for us to increase the grain yield while limiting the environmental impact of agriculture. To achieve these conflicting goals, we must first consider the improvement of both photosynthesis and grain yield for a given crop N content. In this article, I will briefly review photosynthetic performance and yield in rice and wheat in relation to N utilization.

LEAF PHOTOSYNTHESIS

Photosynthesis, Biomass, and Yield in Crops

More than 90% of crop biomass is derived from photosynthetic products. Therefore, many crop scientists have believed that enhancing photosynthesis at the level of the single leaf would increase yields. When C₄ photosynthesis was discovered in 1960s, this expectation rose. For example, the high rates of C₄ photosynthesis in maize (Zea mays) and sugarcane (Saccharum officinarum) were always associated with greater productivity than C₃ crops. On the other hand, a lack of correlation between photosynthesis and plant yield has been frequently observed when different genotypes of a crop are compared (for wheat, Evans and Dunstone, 1970; for rice, Takano and Tsunoda, 1971). This is also true because modern cultivars have been bred for various traits besides photosynthesis. Since many recent studies on elevated [CO₂] experiments show a close relationship between enhanced photosynthesis, biomass, and yield, this suggests that increasing photosynthesis increases yield when other genetic factors are not altered (Long et al., 2006b). In addition, Murata (1981) reviewed the relationship between potential leaf photosynthesis and maximal crop growth rate of many crops and found a highly positive correlation between them. This also indicates that photosynthesis at the single-leaf level can be an important factor for potential biomass production. At the same time, he found that rice and wheat had higher photosynthesis than other C₃ crops.

Difference in Rubisco Properties between Rice and Wheat

Since photosynthetic capacity is closely related to leaf N content, higher photosynthesis in rice and wheat may be the result of breeding for cultivars with higher leaf N content depending on heavy N fertilization. When light-saturated photosynthesis in air is plotted against leaf N content (both expressed per unit of leaf area), however, a great deal of variation exists among C₃ species. Evans (1989) found that rice and wheat show the highest rates of photosynthesis per unit of leaf N content, up to 10 times higher than some evergreen trees (Fig. 1). This variation is also evident in the data when expressed on a dry weight basis. Such higher rates of photosynthesis in both crops may be caused by greater N allocation to Rubisco (Makino et al., 1992) and higher mesophyll conductance (von Caemmerer and Evans, 1991) compared to other plants. Rubisco is the primary CO₂ fixation enzyme, and the amount and kinetic properties of this enzyme strongly affect the photosynthetic rate. In addition, Rubisco has a low rate of catalysis, and therefore a great deal of N is invested in Rubisco protein. However, both the amount and properties of Rubisco and the CO₂ diffusion resistance differ greatly between rice and wheat.

Figure 1.

Light-saturated rate of CO₂ assimilation in air versus leaf N content, both expressed per unit leaf area. ▲, Wheat; ○, rice; black and white squares, Raphanus raphanistrum; ▴, Death valley annuals; □, Illinois annuals; •, Alocasia macrorrhiza; ▪, Lepechinia calycina; ◇, Californian evergreen trees and shrubs and rainforest trees; ▿, South Africa shrubs; ♦, Prunus ilicifolia. Reprint from Evans (1989).

Rice allocates 25% to 30% of leaf N to Rubisco with higher affinity for CO₂ (20% lower K_m for CO₂) than wheat, whereas wheat allocates 20% to 25% of leaf N to Rubisco with greater k_cat (50% higher V_max for carboxylation) than rice (Makino et al., 1988). The CO₂/O₂ specificity for carboxylation and oxygenation does not differ between the two species. The solubility of CO₂ in water decreases at higher temperatures. Coupled with this, the present low atmospheric CO₂ levels enhance photorespiration at elevated temperatures. Therefore, selection pressure for Rubisco with higher affinity for CO₂ may be imposed on rice that originates in warm and hot regions. In contrast, since the CO₂ solubility in water increases at low temperatures, the Rubisco k_cat may be more important in wheat grown in cool environments. Thus, Rubiscos from both species may have evolved to improve plant performance in their typical habitats, respectively. Actually, light-saturated photosynthesis measured at present CO₂ levels is higher in rice above 30°C but higher in wheat below 25°C when both species have similar leaf N concentrations (Nagai and Makino, 2009). Since the difference in k_cat between wheat and rice is greater than the difference in K_m(CO₂) and since Rubisco strongly limits photosynthesis at cool temperatures (Makino and Sage, 2007), at cool temperatures the rate in wheat is considerably higher than in rice.

No variability in the kinetic properties of Rubisco has been found among rice varieties, including old and modern cultivars (Makino et al., 1987). Significant higher k_cat was observed only for a C-type genome variety among Oryza species (Makino et al., 1987), suggesting that Rubisco kinetic properties cannot be targeted for conventional crossbreeding among cultivars. Similarly, in the Triticum genus, the variation in the k_cat for a main cultivar of wheat, T. aestivum, is not significant, whereas there are some differences correlated with the genome constitution in the Triticum and Aegilops species (Evans and Austin, 1986). Additionally, Evans and Austin (1986) found that Rubisco from T. aestivum has a higher k_cat than that from Triticum monococcum. Higher k_cat was associated with possession of the B-type cytoplasm genome, which encodes the large subunit of Rubisco. Although the large subunit is more conserved than the small subunit, the results by Evans and Austin suggested that the RbcL gene can be targeted for an improvement in photosynthesis. Recent technology for plastid transformation, at least with tobacco (Nicotiana tabacum), enables more precise manipulation and replacement of Rubisco. For example, introduction of a gene into the chloroplast linking both the large and small subunits of Rubisco led to successful assembly of a novel holoenzyme (Whitney et al., 2009). Such plastid transformation provides a transplastomic tobacco with a foreign Rubisco with large and small subunits derived from different higher plants. However, plastid transformation techniques have still not been developed for the major crop species including rice and wheat.

Difference in CO₂ Diffusion in a Leaf between Rice and Wheat

Rice and wheat allocate different amounts of N to Rubisco in response to an N supply (Evans, 1989; Makino et al., 1992). With increasing N supply, rice allocates an increasing proportion of N to Rubisco. By contrast, wheat allocates a constant proportion of N to Rubisco. To maintain the balance between the in vivo capacities of Rubisco and other photosynthetic processes, a greater increase in Rubisco content in rice may be required to offset changes in mesophyll conductance (Evans and Terashima, 1988). Mesophyll conductance in leaves is estimated to decrease the stromal CO₂ partial pressure by 30% in general under high irradiance conditions (von Caemmerer and Evans, 1991). Furthermore, as a result of this resistance to CO₂ diffusion, the decrease in stromal CO₂ partial pressure becomes more severe with increasing leaf N content. In wheat, an exponential increase in the carbonic anhydrase activity was observed with increasing leaf N content (Makino et al., 1992). Although there is little direct evidence for contribution of carbonic anhydrase to mesophyll conductance, these differential responses of Rubisco and carbonic anhydrase between rice and wheat might be related to the in vivo balance between Rubisco and other factors limiting photosynthesis. Mesophyll conductance is also affected by membrane permeability as a function of aquaporin density (Hanba et al., 2004). Unfortunately, there is no available data on the response of this factor to leaf N content.

Mesophyll conductance is closely related to the surface area of chloroplasts exposed to the intercellular air spaces (Evans and Loreto, 2000; Terashima et al., 2006). Chloroplast surface area is strongly dependent on the total mesophyll cell area, and both of these are higher in rice than in wheat. For example, chloroplasts including stromules cover more than 95% of the mesophyll cell periphery in rice (Sage and Sage, 2009) whereas the exposed surface area of the chloroplast in wheat was estimated to be 76% (Evans and Loreto, 2000). The ratio of total surface area of mesophyll cell to leaf area in rice ranges from 23 to 44, while the values in wheat are between 8 and 24 (von Caemmerer and Evans, 1991). In rice, both this area ratio and intercellular air spaces increase with increasing N application (Chonan, 1970). Consistent with these anatomical data, direct measurement of mesophyll conductance using carbon isotope discrimination during CO₂ uptake showed that rice had a greater conductance than wheat, which increased with increasing leaf N content (von Caemmerer and Evans, 1991). Additionally, mitochondria and peroxisomes are confined to the interior regions of the cytoplasm and encapsulated by the chloroplasts and stromules in rice, whereas they are pressed against the cell periphery by the large vacuole in wheat (Sage and Sage, 2009). Thus, rice mesophyll cells are specialized to maximize both the CO₂ diffusion into the stroma and the refixation of photorespired CO₂ to thrive in hot environments where photorespiration is stimulated (Sage and Sage, 2009).

Improvements in Photosynthesis

Improving photosynthesis is of greatest agronomic importance as the most plausible route toward enhanced biomass production. Although no genetic differences in Rubisco properties are evident, a great deal of apparently genetically controlled variation in stomatal conductance has been recently observed for a given leaf N content or for a given rate of photosynthesis in rice cultivars (Hirasawa et al., 2010). Such variation may provide a starting point for breeding approaches to improve photosynthesis. Improving photosynthetic adaptation to environmental conditions is also another agronomic goal. For example, the success of the northwards extension of rice cultivation was the result of breeding for cool tolerance (Nishiyama, 1993) and the loss of photoperiod sensitivity (Izawa, 2007). No attempt has been made to adapt potential photosynthesis to cool environments. Photosynthetic efficiency and biomass productivity in rice largely decreased at cool temperatures (Nagai and Makino, 2009), and rice yield in northern regions of Japan is limited by low biomass accumulation in cool climates between spring and early summer. If a wheat-type photosynthetic system could be expressed in rice, canopy carbon gain would be appreciably increased in cool climates such as northern regions in Japan and China. The replacement of wheat Rubisco with a high k_cat form may be effective. As described above, since the photosynthetic system of rice may have evolved to adapt to warm and hot environments, engineering C₄ photosynthesis may not necessarily be suitable in some points. Rice has greater Rubisco content of the low K_m(CO₂)/low k_cat form (Makino et al., 1988). In addition, rice mesophyll cells are specialized to effectively scavenge photorespired CO₂ (Sage and Sage, 2009). These characteristics are incompatible with engineering C₄ photosynthesis. Another important adaptation will be to the steadily elevating atmospheric CO₂ concentration. Elevating CO₂ may be more effective for wheat than for rice because of wheat’s higher K_m(CO₂)/high k_cat form and lower N allocation to Rubisco. Some elevated CO₂ experiments show that stimulation of photosynthesis is greater in wheat than in rice (Long et al., 2006a). Furthermore, Rubisco content in both species exceeds the level necessary for maximal photosynthesis in elevated [CO₂] environments (for rice, Makino et al., 1997; for wheat, Theobald et al., 1998). Since large amounts of N are invested in Rubisco, an attempt to decrease N allocation to Rubisco may lead to the improvement in photosynthesis per unit of leaf N content. In fact, antisense RBCS rice with theoretically optimal Rubisco content at elevated CO₂ concentrations shows higher rates of photosynthesis only under conditions of elevated CO₂ (Makino et al., 1997). This construction may be one of the model crops that perform better under low N input conditions in near future high CO₂ environments.

CROP YIELD

Historical increases in cereal yield have depended on large inputs of N fertilizer. Cereal yield is determined by grain numbers per unit land area, grain weight, and the proportion of grains that fill. In rice, since single grain weight is genetically constant irrespective of N application and growth environments (Yoshida, 1981), yield is simply determined by the product of the two elements: grain number and the ratio of filled grains, both of which are affected by N application. Although a negative correlation between them is frequently observed when the yield target is high (Matsushima, 1993), in many cases rice yield is limited by the grain number (Yoshida, 1981; Yoshida et al., 2006). Therefore, the important issue for achieving a high yield in rice is enhancing the grain number with a high proportion of ripened grains. In wheat, however, since single grain weight varies depending on growth conditions, yield is affected by both the grain number and size (Fisher et al., 1977; Jamieson et al., 1995). Thus, the targets for a high yield may be more complicated in wheat than in rice.

Grain number in both crops is linearly correlated with total plant N content. This is because N is an important resource, both limiting yield and contributing to the determination of grain number (Sinclair and Jamieson, 2006). In wheat, however, since a negative correlation between grain number and size is frequently observed (Fisher et al., 1977), the correlation between grain number and total N accumulation in crop may differ among varieties and growth conditions. On the other hand, the correlation between grain number and total N in rice has been observed to be independent of variety and growth conditions (Wada and Matsushima, 1962; see Fig. 2A). In the study by Wada and Matsushima (1962), N uptake by the plants depending on growth conditions until the time of heading determined the grain number, irrespective of variety. However, since these results were based on japonica rice cultivars and a limited variety of growth conditions, some attention has been paid to investigate whether genotypic differences in the grain number exist for a given N accumulation (Horie et al., 1997). Recently, Yoshida et al. (2006) reported that the grain number per unit of plant N content is clearly higher in semidwarf indica rice genotypes than in japonica rice genotypes. Since semidwarf indica rice genotypes preferentially tend to differentiate grains on secondary and tertiary rachis branches (Yoshida et al., 2006), these traits may be governed by the indica Gn1 allele that increases grain number (Ashikari et al., 2005). The Gn1 allele from a semidwarf indica rice genotype increases grain number by more than 40% in a japonica rice genotype. In addition, the introduction of semidwarf traits into indica rice genotypes has led to a typical success for high yielding. In genotypes with a large grain sink, the introduction of the dwarfing gene means that assimilates formerly needed for stem growth are effectively used in grain development. Therefore, such semidwarf genotypes may show a large increase in harvest index, defined as the ratio of grain mass to total aboveground biomass at harvest (Evans, 1997; Sinclair, 1998).

Figure 2.

Grain numbers, yield (rough rice), and total biomass aboveground versus total crop N content per unit of land area at harvest in japonica rice cultivars. Red circles, Akita 63 in 2000; red triangles, in 2001; red squares, in 2009; blue squares, Yukigesyou in 2000; blue triangles, Toyonishiki in 2001; purple diamonds, Akitakomachi in 2000; green circles, Akita 39 in 2009. The single grain weight of Akita 63 was 35% larger than that of other cultivars. Therefore, Akita 63 showed higher yield for a given crop N content. Some data are reproduced from Mae et al. (2006).

For japonica rice genotypes, a new type of high-yielding and large-grain cultivar, Akita 63 has been released (Mae et al., 2006). While the grain number of this cultivar did not differ from the common japonica cultivars at any given plant N content (Fig. 2A; as pointed out by Wada and Matsushima, 1962), the single grain weight was about 35% larger. Since single grain weight is genetically constant in rice, such a large grain size without reducing the grain number directly enhances the sink capacity, and the amount of N required for achieving a sink capacity necessary for a high yield was less than in the common cultivar. Consequently, this cultivar showed high yield for a given crop N content (Fig. 2B).

For these high-yielding cultivars, however, a decline in the ratio of filled grains is frequently observed when grown with heavy N application (Matsushima, 1993; Mae et al., 2006). This may be because source capacity has not been improved whereas the grain sink has been successfully enlarged with N application. For example, when the relationships between grain number, biomass, and crop N content at harvest are examined, the grain number increases linearly with increasing total crop N content passing through the origin whereas total biomass increases curvilinearly (Fig. 2). The curvilinear relationship in itself is mainly caused by light use limitation due to an excess leaf area that causes self-shading under conditions of high N application. However, the results in Figure 2 indicate that source capacity limits the yield potential of current high-yielding cultivars when the yield target is high. This means that further improvement in sink capacity is no longer effective and improving photosynthesis and biomass production is the only remaining target for any further increase in the yield potential of today’s high-yielding cultivars.

Grain mass in the modern high-yielding rice and wheat has reached about 60% of the total biomass aboveground at harvest (Evans, 1997; Long et al., 2006b; see Akita 63 in Fig. 2, B and C). This is the highest in all cereal crops. Although it is not apparent whether further increase in the harvest index is feasible (Miura et al., 2010), to substantially enhance yield in both crops will be difficult unless source capacity including photosynthesis is improved by genetic engineering. In addition, since improving source capacity could lead to a decrease in the amount of N required for a high yield, it will reduce the environmental impact of agriculture.