Search tips
Search criteria 


Field Crops Res. 2017 August 1; 209: 168–178.
PMCID: PMC5473176

Breeding drought tolerant rice for shallow rainfed ecosystem of eastern India


In shallow rainfed rice agro-ecosystems, drought stress can occur at any growth stage and can cause a significant yield reduction. During recent years, some rice varieties possessing tolerance of reproductive-stage drought stress have recently been developed. Tolerance of vegetative-stage drought stress is also required to improve rice productivity in drought-prone regions. In this study, we evaluated a set of rice breeding lines for their response to a range of different types of vegetative-stage drought stress in order to propose standardized phenotyping protocols for conducting vegetative-stage drought stress screening trials and also to identify genotypes combining tolerance of vegetative- and reproductive-stage drought stress. A soil water potential threshold of −20 kPa during the vegetative stage was identified as the target for effective selection under vegetative stage with grain yield reduction of about 50% compared to irrigated control trials. Genotypes identified as showing high yield under reproductive-stage drought stress were not necessarily the genotypes showing best performance under vegetative-stage drought stress. Genotypes IR72667-16-1-B-B-3, IR78908-126-B-2-B, and IR79970-B-47-1 showed tolerance of both vegetative-stage and reproductive-stage drought stress. For most, the genotypes that were best under vegetative stage drought or even vegetative stage + reproductive stage drought were different from the genotypes that were best under reproductive stage drought. Based on the cultivar superiority measure, IR69515-6-KKN-4-UBN-4-2-1-1-1 and IR78908-126-B-1-B were the stable genotypes (indicated by low Pi) under both irrigated control and severe vegetative stress conditions, genotypes IR83614-203-B and IR78908-80-B-3-B were stable under irrigated control conditions and moderate stress, whereas IR72667-16-1-B-B-3 was stable under both moderate and severe vegetative-stage stress conditions.

Keywords: Rainfed, Drought stress, Vegetative-stage, Reproductive-stage, Soil water potential

1. Introduction

Drought is the most important factor limiting rice productivity in the rainfed rice agro-ecosystem (Huke and Huke, 1997, Pandey and Bhandari, 2009). Climate change is predicted to increase the frequency and severity of drought, which will likely result in increasingly serious constraints to rice production worldwide (Wassmann et al., 2009). Drought can occur at any stage of the rice crop in any year in rainfed areas. Modern rice varieties are highly sensitive to drought stress at seedling, vegetative, and reproductive stages and even mild drought stress can result in a significant yield reduction in rice (O'Toole, 1982, Torres and Henry, 2016). At seedling stage, drought affects crop establishment and seedling survival rates. At vegetative stage, drought reduces leaf formation and tillering, which subsequently reduces the development of panicles per plant, thus causing a yield loss; whereas, at reproductive stage, drought causes a reduction in the number of grains per panicle, increases grain sterility, and reduces grain weight (Pantuwan et al., 2002).

The reproductive stage is recognized as the most critical stage at which drought stress can cause a high yield reduction (Hsiao, 1973), and drought at the vegetative stage was earlier predicted to have a relatively small effect on grain yield in rice (Boonjung and Fukai, 1996). However, it should be noted that these conclusions are based on the effects of drought stress on the rice plant, rather than on which type of drought stress is most frequently occurring in farmers’ fields. Vegetative-stage drought has become a critical factor in reducing rice yield in shallow rainfed environments in recent years because of the late arrival of monsoon rains or long gaps between initial rains (Bunnag and Pongthai, 2013). In recent years, the frequency and intensity of vegetative-stage drought stress have increased in shallow rainfed areas of South and Southeast Asia, particularly in eastern India. Due to less initial rains, farmers fail to accumulate enough water in the field early in the season to prepare land and undertake transplanting. As a result, large areas in shallow rainfed ecosystem are left un-transplanted in years with less initial rainfall. Even when farmers are able to transplant, slow growth, less tillering, and in some cases death of early transplanted seedlings due to vegetative-stage drought stress cause heavy yield losses.

Increased incidence of vegetative-stage drought stress under climate change requires the development of rice varieties that combine tolerance of vegetative-stage and reproductive-stage drought stress in addition to having high yield potential under well-watered conditions. However, the responses to drought at the vegetative stage are different from the responses to drought at the reproductive stage (Kamoshita et al., 2008). Depending on the drought tolerance mechanism, traits that confer tolerance of reproductive-stage drought may not necessarily be effective under vegetative-stage drought stress. Recent research to improve rice yield under reproductive-stage drought stress while maintaining high yield potential (Verulkar et al., 2010, Mandal et al., 2010, Kumar et al., 2008, Kumar et al., 2014) has resulted in the identification of some QTLs that impart increased yield in the reproductive stage (Bernier et al., 2008, Dixit et al., 2015). These efforts have been successful through the development of standardized reproductive stage drought screening protocols that allowed for clear discrimination of drought-tolerant genotypes from drought-susceptible genotypes (Kumar et al., 2008). The timing and severity of drought (mild, moderate, or severe) in relation to the growth stage was appropriately determined so as to capture the genetic variation for reproductive-stage drought tolerance (Kumar et al., 2008).

However, for vegetative-stage drought stress screening, although some information is available on screening protocols (Verulkar and Verma, 2014), guidelines for the severity of drought stress to be imposed have not been very precisely defined. These guidelines are necessary to aid scientists in deciding on the appropriate timing of irrigation for vegetative-stage drought stress screening experiments. The lack of proper screening methods acts as a constraint and often delays the attainment of breeding objectives (De Datta and Seshu, 1982). Developing protocols for effective vegetative-stage drought stress screening in rice will facilitate the identification and development of varieties that are tolerant of vegetative-stage drought. Currently, very few varieties have been characterized to possess high yield potential as well as tolerance to drought at both stages. Selection for yield and yield-attributing traits at the vegetative and reproductive stages in standardized drought screens as well as high yield potential under well-watered conditions may allow breeders to select lines combining tolerance of vegetative- and reproductive-stage drought stress in high-yielding genetic backgrounds.

In the present study, a set of genotypes previously reported to be tolerant of drought at the reproductive stage (Verulkar et. al., 2010; Raman et al., 2012) and some additional genotypes identified to be tolerant to reproductive stage drought were evaluated for tolerance of vegetative-stage drought stress. The genotypes were developed by crossing high-yielding but drought-susceptible rice varieties with drought-tolerant donors and subjected to direct selection for grain yield under reproductive-stage drought and well-watered conditions. The lines were evaluated over three years under different levels of vegetative-stage drought stress with the aim of (i) devising a suitable dry-season vegetative-stage drought stress screening protocol by carefully monitoring soil moisture status over time, (ii) characterizing the degree of yield reduction under moderate and severe vegetative-stage drought stress, and (iii) identifying breeding lines that show tolerance of both vegetative- and reproductive-stage drought stress, in addition to having high yield potential under well-watered conditions.

2. Materials and methods

2.1. Field sites and experimental designs

2.1.1. Vegetative stress screening

Vegetative-stage stress screening trials at the National Rice Research Institute, Cuttack, Odisha, India (20°27′9″N, 85°56′25″E), were conducted during the dry season over three years, from 2008 to 2010. The experimental site had a sandy loam soil. The genotypes tested were advanced breeding lines of 100-120 days’ duration generated using crosses of popular high-yielding rice varieties with a diverse array of donors for drought tolerance developed at IRRI. A total of 210 lines were evaluated (72 in 2008, 75 in 2009, and 63 in 2010) along with popular-varieties IR64, IR36, and MTU1010 used as checks. An alpha-lattice design with three replications was used in all years for each trial. The vegetative stage was considered to begin following the seedling stage (1 month after direct seeding or 10 days after transplanting) and to end when panicle initiation was observed. Two separate vegetative-stage drought trials were planted in 2008 and 2009, which differed in severity due to differences in planting date, time of initiation of drought stress, and/or time of re-watering, and one vegetative-stage drought trial was planted in 2010. A separate irrigated control treatment was also planted in each year of the study in which the water level was maintained at a depth of ~2 cm through flooding irrigation (Table 1).

Table 1
Trial-wise means and stress categorization of the 2008–2010 vegetative stage drought trials at Cuttack.

In 2008 and 2009, the trials were manually dry direct seeded into dry soil at a depth of 2–3 cm and a seeding rate of 60 kg ha−1 with 4–5 seeds per hill. This method gave uniform seedling emergence for all plots within 6-8 days. The seeds were dibbled in three rows of 2.5 m each at 20 cm x 15 cm spacing in 4-m-length rows. In 2010, 17-21-day-old seedlings (3–4 plants per hill) were transplanted at a spacing of 20 cm × 15 cm (Cuttack) in 4-m-length rows in puddled bunded fields. Inorganic NPK fertilizer was applied at the rate of 40-20–20 kg ha−1. P and K were applied as a single basal dose at the time of sowing, whereas N was applied in two equal splits, one at the 15-20-day-old seedling stage after the first weeding and the other at maximum tillering stage. Weeds were controlled by two hand weedings per season. Insect and pest control measures were applied when required.

2.1.2. Reproductive stress screening

A total of 215 of advanced breeding lines of 100–120 days’ maturity duration which included popular varieties as checks were tested from 2008 to 2010 at four sites chosen to represent a wide range of drought-prone shallow rainfed lowland rice production environments in India. At all sites, the trials were planted under irrigated control and reproductive-stage drought stress in alpha lattice designs with three replications. The general information on each trial is presented in (Table 2). List of lines evaluated under vegetative stage and reproductive stage drought stress is mentioned in Table 3.

Table 2
Details of reproductive stage drought experiments conducted during 2008–2010.
Table 3
Released varieties and breeding lines evaluated for two or more years in the vegetative (Cuttack) and reproductive stage drought trials.

Twenty-one-day-old seedlings were transplanted in puddled fields at 20 cm × 20 cm spacing in 4 rows of 4 m length with total plot size of 3.2 m2. Inorganic NPK fertilizer at the rate of 90–60–40 kg ha−1 was applied. P and K were applied as a single basal dose at transplanting, whereas N was applied in three equal splits, at transplanting and at 25 and 50–56 days after transplanting.

In stress trials, N was applied in two splits, at transplanting and at 25 days after transplanting. Weeds were controlled by two hand weeding.

2.2. Soil moisture monitoring and management of drought stress treatments

All vegetative drought stress trials were conducted in upper fields that did not retain standing water. Irrigation was initially applied at 3-4-day intervals and the soil in the stress plots was maintained saturated up to 25 days after germination/sowing (2008 and 2009) or 4 weeks after transplanting (2010). The stress treatment was then initiated by stopping irrigation. As the soil dried, gravimetric soil water content was monitored at 5-day intervals through soil sampling at the depth of 15–30 cm at three different locations in each replication. The fresh soil samples were weighed and then oven-dried and re-weighed to calculate the moisture content within 2–3 days of sampling ((fresh sample wt. − dry sample wt.)/dry sample wt. × 100). Soil water potential at a depth of 30 cm was recorded on alternate days through tensiometer tubes (Soil moisture Equipment Co.) placed randomly in five to six locations in the stress plots. In the stress treatments, piezometers constructed from perforated PVC pipes wrapped with coir rope or fabric (2-m length) were installed (two per stress experiment) after sowing/transplanting to monitor water table depth.

In each season, the drought stress treatment was maintained until the plants began to show stress symptoms such as severe leaf rolling and tip drying, which occurred at 15 and 53 days in 2008, at 17 and 56 days in 2009, and at 31, 7, 7, 7, and 7 days in 2010, respectively (Table 4). The stress treatments were re-watered by flooding the field with water when gravimetric soil moisture was about 12% and the soil water potential was about −15 kPa at a depth of 30 cm and correspondingly when the water table was deeper than 100 cm. The total rainfall recorded during the period of stress was 21.8 mm and 10.2 mm in 2008 and 2010 DS, respectively, with no rainfall in 2009 (Fig. 1). Once the vegetative-stage drought stress treatments finished, irrigation was resumed to allow the plants to recover from the stress. The recovered crop was grown under weekly irrigation until maturity.

Fig. 1
Climate conditions during the 2008, 2009, and 2010 trials in Cuttack, Odisha, India.
Table 4
Irrigation schedule across the three dry seasons under moderate and severe vegetative-stage stress.

2.2.1. Reproductive stress screening

For each drought-stress experiment, an irrigated control (non-stress) trial was planted at each of the locations. In the non-stress experiments, standing water was maintained at each site from transplanting to 10 days before maturity by maintaining the rainwater in the field or by providing supplementary irrigation through a pump as and when required. The reproductive-stage drought-stress experiments (stress, S) were irrigated like the non-stress experiments by keeping standing water up to 28 days after transplanting. Thereafter, the stress fields were drained to allow them to dry and stress to appear. The stress experiments were not provided with any supplemental irrigation after drainage even if the stress was very severe. This has been reflected in the difference in mean yield of stress trials at different sites (Table 2).

2.3. Traits measured

In all trials, the number of days to flowering (DTF) was recorded when panicles in 50% of the plants in each plot emerged and flowered. Plant height (PHT) was recorded on the main culm of five representative plants from ground level to the tip of the panicle before harvesting. Grain yield was measured from a whole-plot area of 3.0 m2, and was adjusted to 14% moisture content. Drought scores and recovery observations were taken following the 0–9 scoring method of the Standard Evaluation System (SES) (IRRI, 1996).

2.4. Data analysis

In every year of the vegetative as well as the reproductive stage drought trials, the set of genotypes not performing well under drought stress was discarded from the entry list of the subsequent trial and was replaced by new genotypes showing promising performances under drought. Across vegetative stress trials at Cuttack, the genotypes of interest were a subset of 50 genotypes grown in two or three years, of which 17 genotypes (set 1, Table 3) were common across all three years. Of the 215 genotypes tested in reproductive stage stress trials, 47 (set 2, Table 3) were tested in two or more years. 33 entries were common between set 1 and set 2. After harvest, each stress treatment was classified for the observed stress intensity as moderately stressed if the yield reduction compared with that of the irrigated control treatment was 30–65%, and severely drought-stressed if the yield reduction was more than 65%, according to Verulkar et al. (2010). The 2008 vegetative severe stress trial of Cuttack was not used included in the analysis since it showed a mean yield reduction of more than 85% compared with that of the irrigated control (Table 1), as such trials usually show low genetic variability, which prevents reliable differentiation between stress-tolerant and susceptible breeding lines.

Single trial analyses in both vegetative and reproductive drought stress trials were first conducted using a mixed model that considered genotype factor as a fixed effect and replicate and block within replicate effects as random effects. The model is


where gi is the effect of the ith genotype, rk the effect of the kth replicate, bkl the effect of the lth block within the kth replicate, and εikl is the error. The analysis was run using the MIXED procedure of SAS (Littell et al., 2006).

At each stress level a conventional combined analysis of variance across environments (site-year cross-classification) of the reproductive stage stress trials is based on the following model.

yijkl = μ + gi + zj + zrjk + zrbjkl + gzij + εikl

where gi is the effect of the ith genotype, zj is the effect of the jth environment, zrjk is the effect of the kth replicate of the of the jth environment, zrbjklis the effect of the lth block within the kth replicate of the jth environment, gzijis the interaction of the ith genotype in the jth environment and εiklis the residual error. All effects except genotypes were set to random.

The cultivar superiority measure, Pi (Lin and Binns, 1988), based on the year-wise genotype predicted means (1), was used to assess stability. Pi is defined as the mean square of the distance between genotype i and the genotype with the maximum response in the individual treatment. Cultivars with the lowest Pi values are considered the most stable. For each stress level (irrigated control and vegetative-stage drought stress), Pi was obtained by the expression


where PiSis the superiority of the ith cultivar at stress level S, (Xij)S is the mean yield of the ith cultivar in the jth year at stress level S, (Mj)S is the maximum mean response obtained among cultivars in the jth year at stress level S, and n is the number of years at stress level S.

3. Results

3.1. Characterization of vegetative-stage drought stress severity

The severity of stress in each trial depended on the planting date in relation to seasonal rainfall, time of initiation of drought stress, rainfall during the stress treatment, soil properties affecting water retention and drainage, and time of re-watering. In the 2008 and 2009, in direct-seeded trials in Cuttack, gravimetric soil moisture (and soil water potential) declined to lower levels in the severe stress treatments (8–9% (−60 kPa)) than in the moderate stress treatments (16–12% (−25 kPa)) due to 38 days later irrigation of the severe stress treatment compared to the moderate stress treatment (Table 4). The water table declined progressively to below the depth of 60 cm for a major part of the stress period in both the moderate stress (80 cm) and severe stress (100 cm) treatments (Fig. 2). In the 2010 transplanted trial, the soil moisture declined to 12% and −14 kPa and the water table reached a depth of 112 cm at 92 DAS (31 days after initiation of the stress treatment; Fig. 2).

Fig. 2
Soil moisture levels in terms of soil water potential and gravimetric water content at depth of 30 cm, and water table depth during the 2008, 2009, and 2010 trials in Cuttack, Odisha, India.

3.2. Genotypic variation for grain yield under vegetative-stage drought stress

The individual trials were grouped into well-watered, moderately stressed, and severely stressed classes based on the mean yield reduction compared with that of the irrigated control. The percent yield reduction under drought compared with that of the irrigated control ranged from 96% in the 2008 severe stress treatment to 54% in the 2009 moderate stress treatment (Table 1). The trial-wise mean yields were highest in the 2009 irrigated control treatment (4554 kg ha−1) and lowest in the 2008 severe stress treatment (99.6 kg ha−1). The mean yields were relatively low in 2008 under all stress levels (Fig. 3). Significant differences in grain yield among lines were observed in all treatments in all years.

Fig. 3
Genotypic variation for grain yield in each treatment applied in Cuttack. Genotypes are listed in order of the mean yield across vegetative-stage drought stress trials.

Among lines that were tested in all three years across all stress levels, three distinct groups of genotypes showed indications of good performance (Table 5): (1) IR72667-16-1-B-B-3, IR78908-126-B-1-B, and IR79970-B-47-1 showed the highest yield across different conditions with yield of 2.5–5.1 t ha−1 under well-watered conditions and 0.9–2.3 t ha−1 under moderate or severe drought stress. Furthermore, genotype means under reproductive-stage drought stress of entries among the set of 50 tested across six sites during the same years (Verulkar et. al., 2010; Kumar et. al., 2012; Raman et al., 2012) showed that (1) IR72667-16-1-B-B-3, IR78908-126-B-2-B, and IR79970-B-47-1 did well under severe reproductive-stage stress conditions (1.5–1.7 t ha−1); (2) IR78908-80-B-3-B, IR83614-203-B, IR78908-193-B-3-B, IR83614-349-B, and IR83614-438-B showed high yield under moderate stress (1.8–2.9 t ha−1) and moderately high yield under irrigated control conditions (2.8–5.7 t ha−1); and (3) IR69515-6-KKN-4-UBN-4-2-1-1-1 and IR79966-B2-52-2 were the highest-yielding lines under severe vegetative stress. Of these, IR69515-6-KKN-4-UBN-4-2-1-1-1 also showed high yield under irrigated control conditions. Based on the cultivar superiority measure, stable genotypes (indicated by low Pi) under both irrigated control conditions and severe vegetative stress were IR69515-6-KKN-4-UBN-4-2-1-1-1 and IR78908-126-B-1-B (Table 6). Genotypes IR83614-203-B and IR78908-80-B-3-B showed low Pi values under irrigated control conditions and moderate stress, whereas IR72667-16-1-B-B-3 was stable under both moderate and severe vegetative-stage stress conditions (Table 6). Scatter plots (Fig. 4) show that under both moderate stress and severe stress the entries nearly maintained their performances across years with r=0.54** between the moderate stress trials and r=0.69** between severe stress trials respectively which indicates the uniformity in the evaluation of stress tolerance across years.

Fig. 4
Scatter plots for grain yield between years each under moderate and severe stress levels.
Table 5
Overall mean of genotypes across environments (location × year) under irrigated control and managed stress conditions under the vegetative stress experiments at Cuttack and reproductive stage stress experiments at four sites during ...
Table 6
Cultivar superiority measure (Pi) of genotypes tested across two to three years under irrigated control and vegetative-stage drought stress conditions in Cuttack. Cultivars with the lowest Pi values are considered the most stable.

4. Discussion

4.1. Monitoring severity of drought stress in vegetative stress trials at cuttack

If a variety’s potential for success is to be assessed realistically, the field growing conditions and crop management methods must simulate those in the target area where in the variety is expected to be grown (Mackill et al., 1996). Therefore, in studies related to drought tolerance phenotyping, the focus should be on careful choice of the testing environment, an effective irrigation management and timing, criteria and evaluation of yield reduction under water deficit conditions. In order to identify sources of drought tolerance, it is necessary to develop screening methods that are simple, reproducible, and predictive of cultivar performance in the target environment (Serraj et al., 2011).

Monitoring the gradual development of drought in the field, its severity, and its duration aid in categorizing cultivars for vegetative-stage drought tolerance. The aim of such screens is to reduce the mean yield of drought stress trials by 50–60% compared with that of irrigated control trials, which can amplify the genetic differences between the drought tolerant and the drought susceptible lines for effective selection to be employed (Fischer et al., 2012) and lines that consistently perform well under drought can be identified. In 2008 DS, with the imposition of stress on the 28th day after emergence (DAE), water table depth began to increase steadily. The moderate stress plots were re-watered 15 days after stress imposition (44 DAG) at which the water table depth reached 42 cms. Thereafter, irrigation was provided at intervals of four to ten days. The trial mean yield of the moderate stress trials was reduced by 71% as compared to the irrigated control (Table 1, Fig. 2). In the severe stress trial in 2008, the stress period was further extended. The severe stress plots were re-irrigated 53 days after stress imposition (79 DAG) when water-table depth stretched to 95 cms. Trial mean yield was reduced by 96% as compared to the irrigated control. The stress imposition steadily diminished the soil moisture levels in both moderate and severe stress trials (Fig. 2). Tensiometers were not installed in the stress plots during this season. In 2009 DS, irrigation was suspended on 35 DAG and 33 DAG in the moderate and severe stress trials respectively. With the inception of stress, the soil moisture retention curve indicated a progressive increase in SMT with a progressive decrease in SWC. The moderate stress plots were re-watered 17 days after the application of stress (52 DAG) when the SWC was about 15% and the corresponding SWT about −10 kPA. The water table depth at this stage was 70 cms. Thereafter the trial was irrigated at regular interval of about ten days. The trial mean yield of the moderate stress trials decreased by 54% as compared to control. The severe stress plots were re-watered 56 days after the stress imposition (82 DAG). The mean yield was reduced by 77% as compared to irrigated control. In 2010, the vegetative stage drought experiment was transplanted in place of direct seeded due to frequent initial rain. This provided opportunity to evaluate the performance of lines under two diverse system of cultivation (direct seeded and transplanted) likely to be practiced by farmers in different years depending upon the rainfall pattern in the season. In 2010 DS water deficit was imposed on the 29th day after transplanting (DAP). When the SWC was about 12% and the corresponding SWT about −14 kPA coinciding with a water table depth of 112 cms (60 DAP, 31 days after stress imposition), the stress plots were re-watered. Thereafter the stress plots were irrigated at regular intervals of seven days until harvest. The yield reduction due to induced water stress was about 59%. Therefore, if tensiometers are installed, it is suggested that the field should be irrigated when SWT = −10–15 kPA at a depth of 30 cm and SWC = 12–15% to get a stress severity appropriate to employ selection at vegetative stage.

Across the different severities of vegetative-stage drought stress trials in this study, the 2009 and 2010 moderate stress treatments in Cuttack were closest to the target screening criterion of reducing grain yield by 50% compared with that of the well-watered control. These results suggest that the field conditions and timing of re-watering were optimum in those trials compared with the 2008 severe stress treatment in Cuttack, which were re-watered later and less frequently and reduced yield by more than the target level. The advantage of achieving the targeted level of vegetative-stage drought stress is also evidenced by the identification of significant differences in grain yield among genotypes in the 2008 moderate, 2009 moderate and severe and 2010 moderate stress treatments in Cuttack,. Yield reduction under stress compared to control in these experiments ranged between 54 and 77 percent. We therefore recommend that a vegetative-stage drought stress treatment should be drained as early as possible after seedling establishment or transplanting (about 1 month after direct seeding or 25–30 days after transplanting) and re-watered minimally as needed to maintain the soil water potential above −20 kPa (ideally between −12 to −15 kPa) as monitored at a depth of 30 cm to achieve yield reduction between 50 and 77 percent under stress compared to control to clearly distinguish between drought tolerant and drought susceptible genotypes at vegetative stage drought. This soil water potential threshold is notably less negative than the recommended soil water potential threshold of −65 kPa for reproductive-stage drought stress screening, reflecting the different time frames and sensitivities of rice at the vegetative and reproductive stage due to lesser presence of roots at depth to access moisture from deeper soil. When the target-level stress is reached, the field should be irrigated to measure the recovery from vegetative-stage drought stress. The number of days required for the soil water potential to reach optimum kPa will be different in wet-season screens or in soils with different water retention properties. Strategies to maximize the degree of drought stress include timing of sowing to coincide with low-rainfall seasons, locating the experiment in an upper field, and installing drainage such as canals around the field. Rainfall, evaporative demand, and the soil characteristics of a site also influence the type of vegetative-stage drought stress applied. Apart from the test environment, factors that play a critical role in devising an appropriate irrigation protocol include stage of crop growth, physical characteristics of the soil, availability of water supply and monitoring soil-water relationships. In 2010, stress treatment was puddled transplanted that experienced little or no rainfall during the stress period. The soil at Cuttack dried to a water potential of −14 kPa over 31 days, (Fig. 2). The volumetric water potential at the end of the 2010 stress period when the water potential was −14 kPa can be estimated to be about 20% (12% gravimetric water content × bulk density of 1.7 g cm−3 as reported by Singh et al. (in review).

Environmental characterization and accurate measurements of plant and/or soil water status are critical in any experiment for which one is concerned about understanding the effects of differing water supply (Jones, 2007). Another important physical characteristic of fields used for vegetative-stage drought stress screening is the uniformity of the experimental field, since variation in the moisture-holding capacity of the soil can decrease the accuracy of the screening (Mackill et al., 1996) and complicate phenotyping. Furthermore, the dry season is useful for preliminary vegetative-stage screening of large numbers of genotypes as it is relatively easier to control soil moisture status (De Datta et al., 1988, Lafitte et al., 2006) and the chances of obtaining a reliable drought screen is expected to be higher as compared to the wet-season screening. However, responses among genotypes may differ from those under wet-season drought, which is typically the main cropping season (De Datta and Seshu, 1982). Nevertheless, the dry season may be useful for preliminary vegetative stage screening of large numbers of genotypes (Pantuwan et al., 2004). The selections from dry season screening can be tested for yield and other agronomic performances in succeeding wet seasons. The actual field drought reactions would be observed during a wet season when dry spells of varying degrees of severity occur (De Datta and Seshu, 1982). Farmers require cultivars that not only respond to favorable conditions but also produce an economically safe yield under drought conditions. Therefore, breeding lines need to be screened under both stress and non-stress conditions. Differences in climatic conditions between seasons can therefore be monitored and the performance of selections from dry-season screening can be validated in succeeding wet seasons if needed.

The evaluation of yield stability by the level of stress can be useful for targeting selection to regions with either a narrow range of stress severities or a high degree of variability in the level of stress across seasons. Previous studies have used Pi for assessing varietal stability and cultivar evaluation (Aremu et al., 2007, Silva et al., 2008, Bastos et al., 2009, Koppel and Ingver, 2010, Fikere et al., 2014). In this study, the identification of genotypes performing well under vegetative-stage drought stress was conducted in reference to their previous performance under reproductive-stage drought stress. Interestingly, some genotypes in this study that had previously been well-documented as high-yielding under reproductive-stage drought stress did not consistently stand out for grain yield under vegetative-stage drought stress. IR74371-70-1-1 and IR55419-04 were earlier identified to be reasonably good yielding and stable under both irrigated (4 and 5 t ha−1, respectively) and reproductive-stage drought stress (2.0–2.2 t ha−1; Verulkar et al., 2010, Kumar et al., 2012a, Kumar et al., 2012b, Raman et al., 2012), but showed only moderate performance under vegetative-stage drought (0.8–1 t ha−1). In comparison, IR72667-16-1-B-B-3, IR78908-126-B-1-B, and IR79970-B-47-1 performed well under both vegetative- and reproductive-stage drought stress, and were the most stable yielding under both moderate and severe vegetative-stage stress. These results indicate the involvement of the different plant traits responsible for resistance at the two stages and the distinct response of rice genotypes to vegetative-stage and reproductive-stage drought stress, but also suggest that some genotypes can show tolerance to both stages of drought stress.

Despite being highly susceptible even to moderate reproductive-stage drought stress, IR36 and IR64 showed good tolerance and yielded 1.6–2.2 t ha−1 under moderate vegetative-stage stress in this study. IR64 and IR36 are high-yielding lowland varieties popular in many rainfed areas of South Asia and Southeast Asia that have good yield potential under favorable conditions and preferred quality characteristics. Although developed for irrigated ecosystem, tolerance to mild to moderate vegetative stage stress could be possible reason for such varieties to become popular in rainfed drought prone ecosystems. However, due to susceptibility to reproductive stage drought stress, they need replacement (Atlin et al., 2008) and the varieties combining tolerance to vegetative and reproductive stage drought stress could help achieve providing appropriate varieties for cultivation to farmers of rained ecosystems. Since the selected lines from the present study showed yield similar to that of IR36 and IR64 under irrigated non-stress as well as possessed tolerance to both vegetative and reproductive stage drought tolerance, they may be appropriate candidate varieties targets as genotypes to replace IR36 and IR64. Among the lines found promising in the present study, IR 72667-16-1-B-B-3 has been released as a drought tolerant variety for rainfed lowland ecosystem in the Philippines.

5. Conclusions

For vegetative-stage drought screening in the dry season, maintaining the soil water potential above −20 kPa as monitored at 30-cm depth is recommended to be able to clearly distinguish between tolerant and susceptible genotypes at the time of harvest. This study identified some genotypes that were susceptible to drought in reproductive-stage drought stress but showed tolerance of vegetative-stage drought stress, indicating the distinct response of rice to drought stress at these two growth stages. Combining tolerance of vegetative-stage drought with tolerance of reproductive-stage drought could be accomplished by performing separate vegetative- and reproductive-stage drought stress screening trials, which in this study resulted in the identification of genotypes IR72667-16-1-B-B-3, IR78908-126-B-2-B, and IR79970-B-47-1 as tolerant of both stages of drought stress. Alternatively, tolerance of both vegetative-stage and reproductive-stage drought stress could be accomplished by crossing donor lines, one of which is tolerant of vegetative-stage drought stress and one of which is tolerant of reproductive-stage drought stress. The development of improved varieties with combined tolerance of drought stress at multiple growth stages will help farmers in rainfed rice-growing regions maintain stable yields across increasingly unpredictable climatic conditions


The present study was supported by a grant from the Bill and Melinda Gates Foundation, USA “Stress tolerant rice for poor farmers of Africa and South Asia”.


Aremu C.O., Ariyo O.J., Adewale B.D. Assessment of selection techniques in genotype x environment interaction in cowpea Vigna unguiculata (L.) Walp. Afr. J. Agric. Sci. 2007;2(8):352–355.
Atlin G., Venuprasad R., Bernier J., Zhao D., Virk P., Kumar A. Rice germplasm development for drought-prone environments: progress made in breeding and genetic analysis at the International Rice Research Institute. In: Serraj J., Bennett J., Hardy B., editors. Drought Frontiers in Rice: Crop Improvement for Increased Rainfed Production. World Scientific Publishing and Los Baños (Philippines): International Rice Research Institute; Singapore: 2008. (400 p)
Bastos I.T., Barbosa M.H.P., Resende M.D.V.D., Pedrozo C.A., Melo C.G., Peternelli L.A., Costa P.M.D.A., Xavier C.V., Baffa D.C.F. Correlation among predicted genotypic values and adaptability and stability estimates of sugarcane clones in a mixed model context. Sci. Agraria. 2009;10(2):111–118.
Bernier J., Atlin G.N., Serraj R., Kumar A., Spaner D. Breeding upland rice for drought resistance. J. Sci. Food Agric. 2008;88:927–939.
Boonjung H., Fukai S. Effects of soil water deficit at different growth stages on rice growth and yield under upland conditions. 2. Phenology, biomass production and yield. Field Crops Res. 1996;48:47–55.
Bunnag S., Pongthai P. Selection of rice (Oryza sativa L.) cultivars tolerant to drought stress at the vegetative stage under field conditions. Am. J. Plant Sci. 2013;4:1701–1708.
De Datta S.K., Seshu D.V. International Rice Research Institute; Los Baños, Laguna Philippines: 1982. Evaluating Rices for Drought Tolerance Using Field Screening and Multilocation Testing. Drought Resistance in Crops with Emphasis on Rice; pp. 245–263.
De Datta S.K., Malabuyoc J.A., Aragon E.L. A field screening technique for evaluating rice germplasm for drought tolerance during the vegetative stage. Field Crops Res. 1988;19:123–134.
Dixit Shalabh, Grondin Alexandre, Lee Cheng-Ruei, Henry Amelia, Olds Thomas-Mitchell, Kumar Arvind. Understanding rice adaptation to varying agro-ecosystems: trait interactions and quantitative trait loci. BMC Genet. 2015;16(86) [PMC free article] [PubMed]
Fikere M., Bing D.J., Tadasse T., Ayana A. Comparison of biometrical methods to describe yield stability in field pea (Pisum sativum L.) under south eastern Ethiopian conditions. Afr. J. Agric. Res. 2014;9(33):2574–2583.
Fischer Ken S., Fukai Shu, Kumar Arvind, Leung Hei, Jongdee Boonrat. Phenotyping rice for adaptation to drought. Front. Physiol. 2012;3:282. [PubMed]
Hsiao T.C. Plant responses to water stress. Ann. Rev. Plant Physiol. 1973;24:519–570.
Huke R.E., Huke E.H. IRRI; 1997. Rice Area By Type Of Culture: South, Southeast, And East China: A Revised Database.
International Rice Research Institute (IRRI) 4th edition. IRRI; The Philippines: 1996. Standard Evaluation System for Rice.
Jones H.G. Monitoring plant and soil water status: established and novel methods revisited and their relevance to studies of drought tolerance. J. Exp. Bot. 2007;58:119–130. [PubMed]
Kamoshita A., Chandra Babu R., Manikanda Boopathi N., Fukai S. Phenotypic and genotypic analysis of drought-resistance traits for development of rice cultivars adapted to rainfed environments. Field Crops Res. 2008;109:1–23.
Koppel R., Ingver A. Stability and predictability of baking quality of winter wheat. Agron. Res. 2010;8:637–644.
Kumar A., Bernier J., Verulkar Satish Lafitte H.R., Atlin G.N. Breeding for drought tolerance: direct selection for yield: response to selection and use of drought-tolerant donors in upland and lowland-adapted populations. Field Crops Res. 2008;107:221–231.
Kumar A., Verulkar S., Mandal N.P., Variar M., Shukla V.D., Dwivedi J.L., Singh B.N., Singh O.N., Swain P., Mall A.K., Robin S., Chandrababu R., Jain A., Haefele S.M., Piepho H.P., Raman A. High-yielding drought-tolerant, stable rice genotypes for the shallow rainfed lowland drought-prone ecosystem. Field Crops Res. 2012;133:37–47.
Kumar A., Verulkar S.B., Mandal N.P., Variar M., Shukla V.D., Dwivedi J.L., Singh B.N., Singh O.N., Swain P., Mall A.K., Robin S., Chandrababu R., Jain A., Haefele S.M., Piepho H.P., Raman A. High-yielding, drought-tolerant, stable rice genotypes for the shallow rainfed lowland drought-prone ecosystem. Field Crops Res. 2012;133:37–47.
Kumar Arvind, Dixit Shalabh, Ram T., Yadaw R.B., Mishra K.K., Mandal N.P. Breeding high-yielding drought-tolerant rice: genetic variations and conventional and molecular approaches. JXB. 2014 [PMC free article] [PubMed]
Lafitte H.R., Li Z.K., Vijayakumar C.H.M., Gao Y.M., Shi Y., Xu J.L., Fu B.Y., Yu S.B., Ali A.J., Domingo J., Maghirang R., Torres R., Mackill D. Field Crops Res. 2006;97(1):77–86.
Lin C.S., Binns M.R. A superiority measure of cultivar performance for cultivar x location data. Can. J. Plant Sci. 1988;68:193–198.
Littell R.C., Milliken G.A., Stroup W.W., Wolfinger R.D., Schabenberger O. 2nd ed. SAS Institute Inc.; Cary, NC: 2006. SAS for Mixed Models.
Mackill D.J., Coffman W.R., Garrity D.P. International Rice Research Institute; Manila, Philippines: 1996. Drought Resistance. In: Rainfed Lowland Rice Improvement; pp. 79–110.
Mandal N.P., Sinha P.K., Variar M., Shukla V.D., Perraju P., Mehta A., Pathak A.R., Dwivedi J.L., Rathi S.P.S., Bhandarkar S., Singh B.N., Singh D.N., Panda S., Mishra N.C., Singh Y.V., Pandya R., Singh M.K., Sanger R.B.S., Bhatt J.C., Sharma R.K., Raman A., Kumar A., Atlin G. Implications of genotype x input interactions in breeding superior genotypes for favorable and unfavorable rainfed upland environments. Field Crops Res. 2010;118:135–144.
O’Toole J.C. Drought Resistance in Crops with Emphasis on Rice. IRRI; Los Baños, Philippines: 1982. Adaptation of rice to drought prone environments. (195-)
Pandey S., Bhandari H. Drought: economic costs and research implications. In: Serraj J., Bennet J., Hardy B., editors. Drought Frontiers in Rice: Crop Improvement for Increased Rainfed Production. World Scientifi c Publishing; Singapore: 2009. pp. 3–17.
Pantuwan G., Fukai S., Cooper M., Rajatasereekul S., O’Toole J.C. Yield response of rice (Oryza sativa L.) genotypes to different types of drought under rainfed lowlands. Part 1. Grain yield and yield components. Field Crops Res. 2002;73:153–168.
Pantuwan G., Fukai S., Cooper M., Rajatasereekul S., O’Toole J.C., Basnayake J. Yield response of rice (Oryza sativa L.) genotypes to drought under rainfed lowlands 4. Vegetative stage screening in the dry season. Field Crops Res. 2004;89:281–297.
Raman A., Verulkar S.B., Mandal N.P., Variar M., Shukla V.D., Dwivedi J.L., Singh B.N., Singh O.N., Swain P., Mall A.K., Robin S., Babu R.C., Jain A., Ram T., Hittalmani S., Haefele S., Piepho H.P., Kumar A. Drought yield index to select high yielding rice lines under different drought stress severities. Rice. 2012;5:31. [PubMed]
Serraj R., McNally K.L., Slamet-Loedin I., Kohli A., Haefele S.M., Atlin G., Kumar A. Drought resistance improvement in rice: an integrated genetic and resource management strategy. Plant Prod. Sci. 2011;14(1):1–14.
Silva F.L.D., Soares P.C., Cargnin A., de Souza M.A., Soares A.A., Cornélio V.M.D.O., Sousa Reis M. Methods of adaptability and stability analysis in irrigated rice genotypes in Minas Gerais, Brazil. Crop Breed. Appl. Biotechnol. 2008;8:119–126.
Singh S.P., Jain A., Anantha M.S., Tripathi S., Sharma S., Kumar S., Prasad A., Sharma B., Karmakar B., Bhattarai R., Das S.P., Singh S.K., Shenoy V., Babu R.C., Robin S., Swain P., Dwivedi J.L., Yadaw R.B., Mandal N.P., Ram T., Mishra K.K., Verulkar S.B., Aditya T., Prasad K., Perraju P., Mahato R.K., Sharma S., Raman K.A., Kumar A., Henry A. 2017. Rainfall, Soil Sand Content, and Depth of Soil Compaction Predominantly Affect Rice Yield Reduction by Reproductive-Stage Drought at Varietal Screening Sites in Bangladesh, India, and Nepal. (In review)
Torres R.O., Henry A. Yield stability of selected rice breeding lines and donors across conditions of mild to moderately severe drought stress. Field Crops Res. 2016 in press.
Verulkar S.B., Verma S.K. Screening protocols in breeding for drought tolerance in rice. Agric. Res. 2014;3(1):32–40.
Verulkar S.B., Mandal N.P., Dwivedi J.L., Singh B.N., Sinha P.K., Mahato R.N., Swain P., Dongre P., Payasi D., Singh O.N., Bose L.K., Robin S., Babu R.C., Senthil S., Jain A., Shashidhar H.E., Hittalmani S., Cruz C. Vera, Paris T., Hijmans R., Raman A., Haefele S., Serraj R., Atlin G., Kumar A. Breeding resilient and productive rice genotypes adapted to drought-prone rainfed ecosystems of India. Field Crops Res. 2010;117:197–208.
Wassmann R., Jagadish S.V.K., Sumfleth K., Pathak H., Howell G., Ismail A., Serraj R., Redona E., Singh R.K., Heuer S. Regional vulnerability of climate change impacts on asian rice production and scope for adaptation. Adv. Agron. 2009;102:91–133.