Effect of Drought on Oryza glaberrima Rice Accessions and Oryza glaberrima Derived-lines

MATERIALS AND METHODS

Genetic materials: A population of 202 backcross inbred lines developed from a cross between CG14 (O. glaberrima) as donor and WAB56-104 (O. sativa subsp. japonica) as recurrent parent, was used in the first experiment. CG14 is adapted to drought conditions and is characterized by low yield potential but several useful traits, including high weed-competitiveness (Dingkuhn et al., 1997) as a result of early vigor and high tiller number. WAB56-104 is moderately drought tolerant (Hounkpatin, 2007) and an improved upland line developed by AfricaRice, with important agronomic traits such as high yield and short growth duration. The two parents were included as checks in the experiments.

In the second experiment, 211 O. glaberrima accessions of rice were evaluated. They originated from seven West African countries: Cote d’Ivoire, Ghana, Liberia, Mali, Nigeria, Senegal and Sierra Leone. The following checks were used in the experiment: (a) three local landraces: PADCKONO, DCKONO and Salikatato, (b) two O. glaberrima (RAM1 and RAM55) as susceptible and tolerant checks, respectively, based on results of several field experiments (Hounkpatin, 2007; Africa Rice Center (AfricaRice) 2010, unpublished) and (c) CG14 and WAB56-104 as additional checks.

The third experiment evaluated 60 rice genotypes made up of introgression lines (90%), their parents (3.3%): female IR64 (O. sativa) and male TOG5681 (O. glaberrima) used as checks and four of NERICA varieties (i.e., 6.7%) adapted to lowland ecology (NERICA-L varieties): NERICA-L 19, NERICA-L 20, NERICA-L 41 and NERICA-L 42. TOG 5681 is characterized by drought tolerant capacity. The four NERICA-L varieties were developed from the same parents: NERICA-L 19 and NERICA-L 20 from TOG5681/3xIR 64 and NERICA-L 41 and NERICA-L 42 from TOG5681/4xIR 64 and were also used in the study as additional checks. The introgression lines or a set of overlapping chromosome segment substitution lines were developed by the Institut de la Recherche pour le Développement (IRD). These lines are different from those used in the first experiment in that the introgression lines were developed using markers. The genetic materials used in the first and third experiments were obtained from the gene bank of AfricaRice, Togoudo, Benin and those used in the second experiment from IRD.

Experimental site: The first and second experiments were conducted in an upland environment and the third in a lowland environment at the AfricaRice research station, Togoudo, southern Benin (25 m above sea level; 6°25'N, 2°20'E). The station is located in the coastal savannah zone, with a sub-equatorial climate. Southern Benin is characterized by a long dry season (December-March), during which rice cannot be grown without supplemental irrigation, as the available soil moisture during this period is below the total evapotranspiration needs of the rice crop. There is also a water shortage, one-month dry season in southern Benin, usually experienced between mid-July and mid-August.

Experimental details: Three experiments were conducted between July 2006 and April 2008. Each of the experiments was repeated once. The first experiment was direct-seeded with three seeds per hill. Plantlets were later thinned to one healthy plant per hill soon after emergence, at a spacing of 0.20 m between and within rows. Plots were 1.0x1.5 m and were arranged in a split-plot design replicated twice, with irrigation regime as the main-plot factor and rice lines as the sub-plot factor. Within each sub-plot, the lines were randomized using an alpha-lattice design. Two irrigation regimes were used: fully irrigated up to maturity and imposition of 21 days of drought (Efisue, 2006) from 45 Days After Sowing (DAS). Plants were sprinkler-irrigated using a pipe with water from a borehole to maintain the soil above field capacity until 45 DAS, when the treatments were applied according to the experimental design. The irrigation continued in fully irrigated plots until the end of the experiments. Compound fertilizer (NPK 15-15-15) was applied at the rate of 200 kg ha^-1 two weeks after sowing, followed by 40 kg N ha^-1 as urea (46% N) 40 DAS.

Similar protocol was used in the second experiment with the differences that the subplots were 1.5x1.5 m, the distance between plants was 25 cm and between subplots, 75 cm. The experimental plots were fertilized with 15 kg NPK (70-40-40) ha^-1 at two and three weeks after sowing, followed by urea application at the rate of 40 kg ha^-1 (18.4 kg N ha^-1). The plots were arranged in a lattice design with four blocks.

In the third experiment, a split-plot design with three replicates in which the blocks were arranged in an Alpha lattice design, was used. Plots were 1x1 m with one genotype each. The distance between two consecutive lines and two consecutive plants on a line was 20 cm. Plots were fertilized with 40-40-40 kg ha^-1 of NPK (15-15-15) and then with urea (30 kg ha^-1 N), 20 DAS. Decis (Deltamethrin) and Mancozeb (Dithiocarbamate) were applied, respectively, as insecticide at 40 mL/15 L of water and as fungicide at 80 g/15 L of water. The three experiments were kept weed-free by regular hand weeding and bird damage was controlled using bird scares. Plants were harvested at maturity (four months after sowing).

Measurements: Data on rainfall, evapotranspiration, temperature and relative humidity during the experimental periods were recorded by the AfricaRice weather section. In all experiments, data were collected on agronomic traits of the plants following the Standard Evaluation System (SES) for rice (INGER and IRRI, 1996). Four plants were randomly selected and labeled in each experimental plot for the measurement of plant height, which was taken as the distance from the soil surface to the tip of the tallest panicle of each plant. Greenness of leaves (SPAD) was recorded using a SPAD meter on the last fully developed leaf. In all cases, the SPAD meter (SPAD-502, Konica Minolta) was placed in the middle of the widest part of the leaf. Leaf temperature was recorded using a handheld infrared thermometer (Model AG-42, Telatemp Corporation, Inc., Fullerton, CA, USA) placed in the middle (the widest part) of the last fully developed leaf. Leaf temperature and SPAD were recorded between 9 and 10 am. The following traits were also recorded: number of tillers per plant; number of days from sowing to exertion of panicles (flowering) in the first plants and in 50 and 100% of the plants in a plot; the grain yield per plant, determined after drying at 50°C for three days, with moisture content adjusted to 14%. Maturity was recorded as the number of days from sowing to grain ripening (defined as the point at which 85% of grains were mature). Leaf tip burning and leaf rolling were evaluated visually in all experiments at the end of the drought treatment both using scales of 0-9.

Statistical analyses: Analysis of variance (ANOVA) and Pearson correlation analysis were performed using SAS (version 9.1) statistical software (SAS Institute, 2003). The Fisher’s Least Significant Difference (LSD) test was used to test the difference between plant tolerance levels. Genotypes were classified into three groups (sensitive, moderately tolerant and tolerant) according to their leaf rolling and leaf tip burning potential using a Principal Component Analysis (PCA).

RESULTS

Rainfall, evapotranspiration, air temperature and relative humidity varied across seasons and therefore, varied across experiments and genotype groups (Table 1). Little rain (mean of 0.8-2.6 mm/day) was recorded during the experiments on introgression lines and second-season O. glaberrima accessions. More rain (mean of 18.6 to 28.1 mm day^-1) was recorded during experiments on interspecific lines and during the first-year experiment on O. glaberrima accessions. The mean daily evapotranspiration during the experiments on the different genotype types ranged between 3.2 and 4.5 mm. Mean air temperature value ranged between 26.0 and 27.7°C and mean relative humidity between 70.6 and 82.7%.

Variations were observed across plant traits and irrigation regimes for each tolerance level (Table 2). The rice genotype types evaluated in this study showed tremendous variation in response to water level. For introgression lines under fully irrigated conditions, values for plant height, flowering (start, 50 and 100%) and maturity were significantly higher (p≤0.05) for sensitive genotypes than for tolerant and moderately tolerant genotypes while values for number of tillers, leaf temperature, grain yield and SPAD were similar across tolerance levels. Under drought conditions, the same five mentioned-above traits, in addition to number of tillers were significantly different for sensitive genotypes compared with the other two types. For interspecific lines under fully irrigated conditions, values for plant height, number of tillers and time to flowering (50%) were significantly higher for sensitive lines than for moderately tolerant and tolerant lines. However, under drought, only plant height and number of tillers were significantly higher for sensitive lines.

Table 1:	Weather data recorded during three experiments conducted at AfricaRice research station, Togoudo, Benin in 2006-2008

Table 2:	Effect of drought on three rice genotype types according to their drought-tolerance levels evaluated under field conditions at AfricaRice, Togoudo, Benin, 2006-2008
*Difference between values for fully irrigated and drought conditions, SD: Standard deviation. For a given plant trait, values with different lower-case letters are significantly different according to the Fishers’ LSD test, Values with different capital letters for the same trait for a given genotype type indicate significant differences between trait values under fully irrigated and drought conditions

In both irrigation regimes, values for leaf temperature, flowering (start and 100%), maturity, grain yield and SPAD were similar across tolerance levels. For O. glaberrima lines, leaf temperature and SPAD were significantly higher for sensitive genotypes than for other genotype types under fully irrigated and drought conditions. Under drought, sensitive lines also matured later. However, under both irrigation regimes, plant height and number of tillers were significantly higher for tolerant lines while only flowering was similar across the genotype’s drought-tolerance levels.

Plant height and grain yield were lower under drought than under fully irrigated condition during the three experiments for all plant tolerance types. The reduction of plant height under drought ranged between 4.3 and 20.9 cm according to the genotype type and the genotypes’ drought-tolerance level. Greatest grain yield reduction due to drought was observed for introgression lines (972.8-1700.8 kg ha^-1) while the lowest yield reduction was noted for interspecific lines (0.5-1.9 kg ha^-1). Flowering (start, 50 and 100%) and maturity were retarded under drought as compared to under full irrigation for introgression lines, interspecific lines and O. glaberrima accessions. This was observed for all drought-tolerance levels except for 50% flowering for interspecific lines with the moderately tolerant level. Mean delay in start of flowering ranged between 0.2 and 8.0 days. Fifty percent of plants in a plot flowered within a delay of 0.8-21.8 days due to drought. The mean difference between flowering (100%) under drought and fully irrigated conditions ranged from 2.9-9.4 days. Maturity was retarded by 1.3-9.7 days under drought as compared to fully irrigated condition. The number of tillers was not significantly affected by drought for any genotype type except for tolerant interspecific lines with the tolerant level. Leaf temperature of introgression and interspecific lines was not affected by drought; however, it was higher under drought than under full irrigation for O. glaberrima accessions, with an increase of 1.8-3.5° under drought as compared to fully irrigated condition.

The relationship between leaf rolling and leaf tip burning and grain yield under drought was strongest for moderately tolerant introgression lines for leaf rolling and leaf tip burning and also for tolerant O. glaberrima for leaf rolling (Table 3). Positive relationship was observed between rolling and grain yield under drought for all genotype types, excluding the moderately tolerant O. glaberrima accessions. The relationship between leaf tip burning and grain yield under drought was negative for all genotype types even though not always significantly.

Some interspecific lines had better grain yield performance than the parents WAB56-104 and CG14 (Fig. 1a). For example, grain yields of lines 151-3-8 and 104-3-5 were higher under both drought and fully irrigated conditions than those of parent WAB56-104. Similarly, grain yields of 116-2-4 and 117-2-6 were higher than that of CG14 under drought and under both drought and fully irrigated conditions, respectively.

Table 3:	Relationship between leaf rolling and leaf tip burning ability and grain yield under drought of rice genotypes evaluated under field conditions at AfricaRice, Togoudo, Benin, 2006-2008

Fig. 1(a-c):	Grain yield of three genotype types, (a) interspecific lines (202) adapted to upland conditions, (b) introgression lines (60) made for lowland conditions and (c) O. glaberrima (211 accessions)-grown under fully irrigated and drought conditions at AfricaRice, Togoudo, Benin, 2006-2008

Grain yield of NERICA-L20, NERICA-L41, NERICA-L42, MPL-15-3, MPL-202-3, SEN-L21-2, SEN-L-10-1, SEN-L17-2 and SEN-26-3 were higher than those of the two checks TOG5681 and IR64 (Fig. 1b). The genotypes TOG5691, TOG6679 and TOG5591 gave grain yields higher under drought and irrigation conditions than the checks CG14, WAB56-104, RAM1 and RAM55 (Fig. 1c).

DISCUSSION

In this study, more weight was given to plant height, number of tillers, leaf temperature, flowering, maturity, grain yield and SPAD to evaluate drought tolerance. Although, some of the drought evaluation methods based on plant body symptoms such as plant wilting, leaf rolling and yield loss are not always appropriate for varietal screening because they are labor intensive (Hirayama et al., 2006), visible plant damage coupled with phenology are, however, believed to be a good indicator of the stress (Talebi et al., 2009; Gathu et al., 2012). Three field experiments were conducted to evaluate the drought effect on various rice genotypes. The results were consistent across experiments (with a few exceptions) for plant height, flowering (start, 50 and 100%), maturity, grain yield and SPAD. The effects on the number of tillers and leaf temperature were, however, not consistent across the experiments.

Plant height and grain yield were reduced under drought as extensively reported earlier by Lafitte et al. (2002b, 2004), Srinivasan (2005), Xu et al. (2005), Bernier et al. (2007), Ikeda et al. (2007), Ganapathy et al. (2007) and Liu et al. (2008). Under drought, plant development is reduced as a consequence of (a) poor root development; (b) reduced leaf-surface traits (form, shape, composition of cuticular and epicuticular wax, leaf pubescence and leaf color), which affect the radiation load on the leaf canopy; (c) delay in or reduced rate of normal plant senescence as it approaches maturity and (d) inhibition of stem reserves (Blum, 2002). The negative effect of drought on plant height and grain yield concurs with the results of previous studies (Efisue, 2006; Hounkpatin, 2007; Ndjiondjop et al., 2010). In this study, the observation was consistent across experiments, indicating that the traits are good indicators that could be used in drought screening tests as claimed by several authors (Talebi et al., 2009; Mitra, 2001; Sangtarash, 2010) that grain yield alone is not a good measure of drought tolerance even though it could apply to drought escape. This assertion broadly confirmed that the intensity of drought effect on both traits varied with the genetic materials (Loresto et al., 1976; Chang et al., 1982; Mackill et al., 1996; Ali, 2011).

The study also indicated strong relationship between grain yield under drought and leaf rolling and leaf tip burning for moderately tolerant introgression lines. Likewise, the same trend was observed between grain yield under drought and leaf rolling for tolerant O. glaberrima. Similar results have been reported by Ali et al. (2011) on maize, Acuna et al. (2008) on rice and Talebi et al. (2009) on wheat. This explains the role of yield increase of leaf rolling and leaf tip burning potential of genotype. Indeed, it has been demonstrated that photosynthesis rate and productivity is majorly critical in dry weight matter and subsequently in yield production (Poshtmasari et al., 2007; Cha-Um et al., 2007). Therefore, genotypes with enrolled or tip burned leaf forage less for light, close their stomata to maintain their inner moisture content and consequently their photosynthetic rate and productivity decrease (Gollan et al., 1986; Turner, 1986).

The delay in flowering under drought is a consequence of a reduction in plant dry-matter production and a delay in panicle exertion (Murty and Ramakrishnayya, 1982). Novero et al. (1985) reported that the delay in flowering depends on the intensity, time and period of drought. Wopereis et al. (1996) observed longer flowering delay when drought occurred during early tillering than when it occurred in mid-tillering stage. Pantuwan et al. (2002) made similar observations and concluded that under prolonged drought, flowering time is an important determinant of rice grain yield. The maturation stage, which is regarded as the period between anthesis and harvest, is also delayed as a result of delayed flowering or when drought appears after flowering. The delays in flowering and maturity could also be considered as good indicators in drought screening tests since the effect of drought on the traits was consistent. The differences observed among genotypes in the delays might be a result of differences in plant water status in the genotypes during the drought and consequently in the drought escape and avoidance potential of the genotypes. Audebert et al. (2002) stated that drought tolerance is complex, involving several interacting physiological, phenological and morphological mechanisms for escape, avoidance, resistance and recovery. They stated that, depending on the drought type, drought tolerance may be controlled by simple plant traits that can lead to simple and cheap breeding programs. To minimize the effect of drought on rice production, drought-tolerant genotypes will therefore be an appropriate farmer-friendly, seed-based technology that will be easy to disseminate.

The study identified genotypes 151-3-8, 104-3-5, 116-2-4, 117-2-6, MPL-15-3, MPL-202-3, SEN-L21-2, SENL-10-1, SENL-17-2, SENL-26-3, TOG5691, TOG6679 and TOG5591 as with grain yield higher than that of the parents and checks used in the experiments under the two irrigation conditions. These genotypes could serve for the aforementioned purpose and could also serve as potential donors and could be introduced into large scale evaluation programs aiming at designing drought tolerant lines. In addition, genotypes MPL-15-3 and MPL-202-3 could be used for further drought tolerant gene fine mapping and gene identification.

The results, obtained on the number of tillers and the leaf temperature, were not consistent across the experiments. With the exception of the tolerant interspecific lines, drought stress did not impact the number of tiller. However, this negative effect of drought, observed on the number of tillers, corroborates results of Efisue (2006), Mostajeran and Rahimi-Eichi (2009) and Ndjiondjop et al. (2010), who explained the phenomenon as a result of poor plant development. Apart from O. glaberrima accessions, which leaf temperature increased under drought condition, leaf temperature was surprisingly not significantly different for both interspecific and introgression lines regardless of the tolerance level and the water regime. The increase of leaf temperature under drought (in some genotype types) concurs with Ingram et al. (1990), who observed a significant correlation between leaf temperature and grain yield under drought. Leaf temperature has been proposed as a good indicator of plant stress level for its association with plant water status under drought (Serraj et al., 2009). Blum et al. (1978), Ishiahara and Saito (1983) and O’Toole et al. (1984) justified this assertion by the rise in the leaf temperature associated with the decrease of the transpiration rate due to the closure of stomata before the occurrence of macroscopical damage. As about 2470 Joules are lost by the plants following transpiration of every gram of water, transpiration tends to cool the plant leaf; the water losses due to transpiration increase with increase in leaf surface area (Arraudeau, 1998). Maurya and O’Toole (1986) reported leaf temperature as a good drought-avoidance indicator, based on the principle of the cooling effect of transpiration, as high transpiration rate occurs in the leaf at lower temperature in genotypes that avoid drought.

CONCLUSION

In this study, O. glaberrima accessions were confirmed to provide useful genes for drought tolerance as reported for most biotic and abiotic stresses of rice. Once these genes are introgressed into high yielding promising lines, these lines showed improved yield and better resistant to drought. Overall, drought stress impacted significantly the traits studied of some genotypes and some of them revealed tolerance to drought, suggesting the genetic variability for drought tolerance in the investigated material. Therefore, based on this limited sample and environments, testing and selection under non-stress and stress conditions alone may not be the most effective for increasing yield under drought stress. As drought is clearly not a single and homogeneous stress, the type of drought must be defined to avoid that drought levels and timing used in a stress nursery identify lines with specific advantageous traits that might not be valuable in other drought scenarios.

ACKNOWLEDGMENTS

The research was supported by the Generation Challenge Programme project lead by the first author and AfricaRice. The authors thank Dr. Hugues Baimey for the critical reading of the manuscript and IRD for providing part of the material.